On The Homotopy Perturbation Method and Its Applications: Farah Lateef Joey

Republic of Iraq
Ministry of Higher Education

and Scientific Research
Al-Nahrain University
College of Science
Department of Mathematics
and Computer Applications
On the Homotopy Perturbation

Method and its Applications
A Thesis
Submitted to the College of Science of Al-Nahrain University in
Partial Fulfillment of the Requirements for the Degree of Master
of Science in Mathematics
By
Farah Lateef Joey
(B.Sc. Math, Al-Nahrain University, 2008)
Supervised by
Asst. Prof. Dr. Ahlam J. Khaleel Lect. Dr.Shatha Ahmad Aziz
Rabee Al-Thanee 1432

March 2011
‫َﻷ‬
‫ﻴﻢ‬
‫)‪(133‬‬
‫اﻹھﺪاء‬
‫…‬
‫(‪.‬‬ ‫)‬
‫اﻟﻰ ﻣﺜﻠﻲ اﻻﻋﻠﻰ اﻟﺬي ﻛﺎن ﯾﺸﺠﻌﻨﻲ دوﻣﺎ‬

‫واﻟﺪي اﻟﺸﮭﯿﺪ‬ ‫ﻋﻠﻰ اﻟﺪراﺳﺔ ﺣﺘﻰ وﺻﻠﺖ اﻟﻰ ھﺬه اﻟﻤﺮﺣﻠﺔ‬
‫اﻟﻰ ﺷﻤﺲ اﺷﺮﻗﺖ ﻓﻲ وﺟﮭﻲ وﺗﺪﻓﻌﻨﻲ‬

‫أﻣﻲ وأﺳﺮﺗﻲ‬ ‫ﻟﻼﻣﺎم دوﻣﺎ‬
‫واﻟﻰ ﻛﻞ ﻣﻦ ﯾﻌﺘﺒﺮ اﻟﺮﯾﺎﺿﯿﺎت ﻃﺮﯾﻘﺎً ﻓﻲ اﻟﺤﯿــــﺎة‬

Acknowledgements
Before anything …
Thanks to Allah for helping me to complete my thesis

I wish to express my deep appreciation and sincere thanks to my
supervisors Dr. Ahlam Jameel and Dr. Shatha Ahmad for their appreciable
advices, important comments support and encouragement during the
research.
It is of my pleasure to record my respect to the staff members of
Mathematics and Computer Applications, whom gave me all facilities during
my work.
Thanks are extended to the College of Science of Al-Nahrain
University for giving me the chance to complete my postgraduate study.
Also, I would like to express my deepest thanks to all my family and
friends, and every one who helped me in my project.
.
Farah
March 2011
Abstract
The aim of this work is to use the homotopy perturbation method to solve special
types of the local and nonlocal problems. This study including the following aspects:
(1) Give some basic concepts of the homotopy perturbation method.
(2) Use the homotopy perturbation method to solve some types of differential, integral
and integro-differential equations.
(3) Describe some nonlocal problems and use the homotopy perturbation method to solve
them.
(4) Use the homotopy perturbation method to solve some real life applications and these
applications are advection-diffusion problems, gas dynamics problem and the ground
-water level problem.

Contents
Abstract……………………………………………………………
Introduction..………………………..…………………………... I
Chapter One: The Homotopy Perturbation Method for
Solving the Ordinary Differential Equations
Introduction …………………………………………………….......1
1.1 Some Basic Concepts of the Homotopy Perturbation Method ….……..1
1.2 The Homotopy Perturbation Method for Solving the Non-linear
Ordinary Differential Equations ………..............................................8
Chapter Two: The Homotopy Perturbation Method for

Solving the Integral and Integro-Differential
Equations
Introduction ………………………………………………………17
2.1 The Homotopy Perturbation Method for Solving Linear Integral
Equations ………...………..…………………………………...17
2.2 The Homotopy Perturbation Method for Solving Non-linear Integral
Equations ………….……………..…………………………….30
2.3 The Homotopy Perturbation Method for Solving Non- linear Fredholm
Integro-differential Equations…...……………………….............35
Chapter Three: The Homotopy Perturbation Method for
Solving Some Nonlocal Problems
Introduction ………………………………………………………47
3.1 Solutions of One-Dimensional Wave Equation with Non- Homogeneous
Neumann and NonLocal Conditions….………………………….48
3.2 Solutions of the Hyperbolic Integro-Differentials Equations with Non-
Homogeneous Neumann and NonLocal Conditions….…………….57
Chapter Four: Solutions of Some Real Life Applications Via

the Homotopy Perturbation Method
Introduction ……………………….……………………………...64
4.1 Advection-Diffusion problems .…………………………………...64
4.2 Gas Dynamics problem ………..…….…………………………...69
4.3 The Ground- Water Level problem .…..…………………………..71
Conclusions and Recommendations………………………….76

References ....…………………………………………………...77
Introduction
The homotopy perturbation method was first proposed by He J. Huan in 1999, where
the solution of this method is considered as the sum of an infinite series which is very rapidly
converge to the accurate solution, [13].
The homotopy perturbation method, presents some advantages: obtaining exact
solutions with high accuracy, minimal calculations without loss of physical verification. This
method has found application in different fields of nonlinear equations such as fluid
mechanics and heat transfer, [10].
Many authors and researchers studied the homotopy perturbation method, say, He J. in
1999, used the homotopy perturbation method for solving nonlinear ordinary differential
equations of the first and second orders, [13], He J. in 2003, solved the Nonlinear ordinary
differential equations with nth order, [14], He J. in 2004, solved the oscillators equation with
discontinuities via the homotopy perturbation method, [17], He J. in 2005, studied the
homotopy perturbation method for solving one dimensional nonlinear wave equation, [18],
Li-Na Z. and He J. in 2006, solved the electrostatic potential differential equation, [19].
Yu-Xi Wang and et al. in 2008, used the homotopy perturbation method for solving
reaction-diffusion equation, [38], Fatemeh S. and Mehdi D. in 2008, solved the deley
differential equations via the homotopy perturbation method, [9], Jafari H. and et al. in 2008,
used the homotopy perturbation method for solving Gas Dynamics Equation, [22], Jazbi B.
I
and Moini M. in 2008, used the homotopy perturbation method for solving
Schrodinger equation, [23].
Lin Jin in 2008, studied the homotopy perturbation method to solve the three
dimensions parabolic and hyperbolic partial differential equations with variable coefficients,
[29], Behrouz R., in 2009, solved nonlinear Volttera partial integro-differential equations of
the second kind, [4], Ghorbali A. R., and et al. in 2009, used the homotopy perturbation
method for solving the three dimensional heat equation with variable coefficients, [11].
Mashallah M. and Mohammad S. in 2010, studied the homotopy perturbation method
for solving linear fuzzy Fedholem integral equations, [33], Allahviranloo T. and et al. in
2010, solved the linear fuzzy Volttera integral equations of the second kind, [1], Roozi A. and
et al. in 2011, studied the homotopy perturbation method for solving nonlinear parabolic and
hyperbolic partial differential equations of one and two dimensions, [35].
The nonlocal problems plays an important role in real life applications and they used in
various field of mathematical physics and in other fields, [3].
Many authors and researchers studied the nonlocal problem, say, Karakostas G. L. and
Tsamatas P. Ch. In 2000, studied a nonlocal boundary value problem for a second order
ordinary differential equations, [25], Beilin S. in 2002, studied the existence of Solutions for
One-Dimensional Wave equations with Nonlocal conditions, [6], Ruyun M. in 2007,
presented a servay of recent results on the existence and multiplicity of solutions of nonlocal
boundary value problem involving second order ordinary differential equations, [36].
II
The purpose of this thesis is to give a full information for the homotopy perturbation
method and its applications for solving the non-linear Fredholm integral and integro-
differential equations. Also, this method is used to solve special types of nonlocal problems.
Moreover, the solutions of some real life applications are obtained via the homotopy
perturbation method.
This thesis consists of four chapters:
In chapter one, some basic concepts of the homotopy perturbation method are
described and used to solve the nonlinear ordinary differential equations with and without
initial conditions.
In chapter two, the solutions of the linear integral equation of the second kind are
obtained via the homotopy perturbation method and its convergence is presented. Also, the
solution of the non-linear integral and integro-differential equations of the second kind are
obtained by means of the homotopy perturbation method.
In chapter three, the homotopy perturbation method is used to solve the one
dimensional wave and hyperbolic integro-differential equations with non-homogeneous
Neumann and nonlocal conditions, respectively.
In chapter four, the solutions of the advection-diffusion problem with initial boundary
conditions, gas dynamics problem with initial condition and the ground water level problem
with non-homogeneous Dirichlet and nonlocal conditions are obtained via the homotopy
perturbation method.
III
Chapter One
The Homotopy Perturbation
Method for Solving the Ordinary
Differential Equations
Chapter One The Homotopy Perturbation Method for Solving the Ordinary Differential Equations
Introduction:
The homotopy perturbation method proposed first by He J. Huan in 1999 for solving
differential and integral equations, linear and nonlinear has been the subject of extensive
analytic and numerical studies. This method has a significant advantage in that it provides an
approximated solution to a wide range of nonlinear problems in applied sciences. In this
method, the solution is considered as the summation of an infinite series which usually
converges rapidly to the solutions, [13].
In this chapter, some basic ideas of this method has been explained.
This chapter consists of two sections:
In section one, some basic concepts of the homotopy perturbation method are
described.
In section two, we use this method for solving the non-linear ordinary differential
equations.
1.1 Some Basic Concepts of the Homotopy Perturbation Method:
In this section, we give some basic concepts of the homotopy perturbation method. To
do this, we recall the following definition:
Definition (1.1.1), [32]:
Let X and Y be two topological spaces. Two continuous functions f : X 

 Y and
-1-
 Y are said to be homotopic, denoted by f  g , if there exists a continuous
g:X 
function H : X  [0,1] 
 Y , such that:
H ( x,0)  f ( x), x  X
H ( x,1)  g ( x), x  X
In this case, H is said to be a homotopy.
Now, to illustrate this definition, consider the following examples:
Example (1.1.2):
Let X and Y be any topological spaces, f be the identity function and g be the zero
function, then define H : X  [0, 1] 

 Y by:
H ( x, p )  x(1  p ), x  X , p  [0,1]
Then H is a continuous function and
H ( x,0)  x  f ( x), x  X
H ( x,1)  0  g ( x), x  X
Therefore f  g .
Next, the following proposition appeared in [32] without proof, here we give its proof.
Proposition (1.1.3):
On the continuous functions  is an equivalence relation.

-2-
Proof:
Let f : X 
 Y be a continuous function, then define H : X  [0,1] 
 Y by:
H ( x, p )  f ( x), x  X , p  [0,1]
Therefore
H ( x,0)  f ( x), x  X
H ( x,1)  f ( x), x  X
and this implies that f  f . Therefore  is a reflexive relation.
To prove  is a symmetric relation, let f  g , then there exists a continuous function
H : X  [0,1] 
 Y such that:
H ( x,0)  f ( x), x  X
H ( x,1)  g ( x), x  X
Define K : X  [0,1] 
Y by:
K ( x, p )  H ( x,1  p ), x  X , p  [0,1]
Then
K ( x,0)  H ( x,1)  g ( x), x  X
K ( x,1)  H ( x,0)  f ( x), x  X
Hence g  f .
To prove  is a transitive relation, let f  g and g  w, then there exist continuous
functions H : X  [0,1] 
 Y and K : X  [0,1] 
 Y
such that:
-3-
H ( x,0)  f ( x), x  X
H ( x,1)  g ( x), x  X
K ( x,0)  g ( x), x  X
K ( x,1)  w( x), x  X
Define L : X  [0,1] 
 Y by:
 f ( x), p0

L( x, p )   H ( x, p )  K ( x, p )  g ( x), 0  p  1
w( x), p 1

Therefore
L( x,0)  f ( x), x  X ,
L( x,1)  w( x), x  X
lim L( x, p )  lim H ( x, p )  K ( x, p )  g ( x)

p 0  p 0
 H ( x ,0 )  K ( x ,0 )  g ( x )
 f ( x)  g ( x)  g ( x)
 f ( x)
 L( x,0), x  X,
and
lim L( x, p )  lim H ( x, p )  K ( x, p )  g ( x)

p 1 p 1
 H ( x,1)  K ( x,1)  g ( x)
 g ( x)  w( x)  g ( x)
 w( x)
 L( x,1), x  X,
Hence L is a continuous function. Therefore f  w. Hence  is an equivalence relation on
the set of all continuous functions.

-4-
Remark (1.1.4):
Let X and Y be two topological spaces, let f :  

  and g :  
  be
continuous functions. Define H :   [0, 1] 

  by:
H ( x, p )  (1  p ) f ( x )  pg ( x ), x  X , p  [0,1]
Then
H ( x,0)  f ( x), x  X
and
H ( x,1)  g ( x), x  X
Therefore f  g .
Let X and Y be two topological spaces and f : X 

 Y be a continuous function. The
equivalence class of f , denoted by [ f ] is defined by:
 Y be a continuous function and f  g 

[f ] g g: X 
and it is said to be a homotopy class of functions of f .
Remark (1.1.6):
(1) It is clear that f  [ f ], for every continuing function f defined from a topological space
X into a topological space Y. Therefore [ f ]   .
-5-
(2) By using theorem (2.4.4) in [31], the set of equivalence classes of  form a partition of
continuous functions.
Let X and Y be two topological spaces. Two continuous functions f and g are said
homotopic relative to A  X if there exists a continuous function H : X  [0,1] 

 Y such
that:
H ( x,0)  f ( x), x  X
H ( x,1)  g ( x), x  X
H (a, p )  f (a )  g (a ), p  [0,1], a  A
Now, to illustrate the basic idea of the homotopy perturbation method, we consider the
following non-linear equation:
A(u )  f ( x), x   (1.1)
where A is any operator, f is a known function of x. The operator A can generally speaking
be divided into two parts L and N, where L is a linear operator, and N is a non-linear
operator. Therefore equation (1.1) can be rewritten as follows:
L(u )  N (u )  f ( x)  0
According to [13], we can construct a homotopy v :   [0,1] 

  which satisfies the
homotopy equation:
H (v, p)  (1  p)[ L(v)  L(u 0 )]  p[ A(v)  f ( x)]  0

-6-
or
H (v, p )  L(v( x))  L(u 0 ( x))  pL(u 0 ( x))  p[ N (v( x))  f ( x)]  0 (1.2)
where p [0,1],  represents the set of all real numbers and u 0 is an initial approximation of
the solution of equation (1.1).
Obviously, from equation (1.2) we have:
H (v,0)  L(v)  L(u 0 )  0
H (v,1)  A(v)  f ( x)  0
The changing process of p from zero to unity is just that of v(x,p) from u 0 (x) to u(x).
Therefore
L(v)  L(u 0 )  A(v)  f ( x), x  
and
u 0 ( x)  u ( x), x  .
Assume that the solution of equation (1.1) can be written as a power series in p as follows:

v ( x, p )   p i v i ( x ) (1.3)
i 0
By setting p=1 in equation (1.3), one can get:

u ( x)  lim v( x, p )   vi ( x) (1.4)
p 1 i 0
which is the solution of equation (1.1)
The series (1.4) is convergent for some cases, [13].
-7-
1.2 The Homotopy Perturbation Method for Solving The Non-Linear Ordinary Differential
Equations, [13]:
In this section, we use the homotopy perturbation method to solve the non-linear ordinary
differential equations with or without (initial or boundary) conditions. To do this, consider the
following non-linear ordinary differential equation:
A( y )  f ( x) , x Ω (1.5)
where A is any differential operator, f is a known function of x. The operator A can be
divided into two parts L and N where L is a linear operator while N is a non-linear operator.
Therefore equation (1.5) can be rewritten as:
L( y )  N ( y )  f ( x)  0, x Ω
So, by the homotopy perturbation method, we can construct a homotopy u :   [0,1] 

 
which satisfies:
H (u , p )  L(u ( x, p ))  L( y0 ( x))  pL( y0 ( x))  p[ N (u ( x, p ))  f ( x)]  0 (1.6)
where p [0,1],  represents the set of all real numbers and y0 is the initial approximation
for the solution of equation (1.5) which satisfies the initial or boundary conditions if they
exist.
Obviously, from equation (1.6) one can have:
H (u ,0)  L(u ( x,0))  L( y0 ( x))  0
H (u ,1)  L(u ( x,1))  N (u ( x,1))  f ( x)  0
The changing process of p from zero to unity is just that of u(x,p) from y0 ( x) to y (x) .
Therefore
-8-
L(u )  L( y0 )  A(u )  f ( x), x  
and
y0 ( x)  y ( x), x  .
Assume that, the solution of equation (1.5) can be written as power series in p:

u ( x, p )   p i u i ( x ) (1.7)
i 0
where u 0 , u1 , ... are the unknown functions that must be determined. By setting p=1 in the
above equation one can obtain:

y ( x)  lim u ( x, p)   ui ( x) (1.8)
p 1 i 0
which is the solution of the differential equation (1.5).
The infinite series given by equation (1.8) is convergent for some cases. However, the
convergence rate depends on the non-linear operator A. The following opinions are suggested
by He J. to ensure that the convergence of the infinite series given by equation (1.8):
1. The second derivative of N(u) with respect to u must be small.
N
2. The norm of L1 must be smaller than one.
u
To illustrate this method, consider the following examples.
Example (1.2.1):
Consider the first order nonlinear ordinary differential equation:
y ( x)  y 2 ( x)  0, x  1 (1.9)
-9-
Here A( y )  y   y and f ( x)  0 . The operator A can be divided into two parts L and N,
2
where L( y )  y  and N ( y )  y 2 .
In this case, equation (1.6) becomes:
u ( x)  y0 ( x)  py0 ( x)  p[u 2 ( x)]  0, p  [0,1]
Assume the solution of the above equation can be written as given in equation (1.7). By
substituting this solution into the above equation one can have:
2
p u i ( x )  y 0 ( x )  py 0 ( x )  p  p i u i ( x )   0

 

i 0
i
  
 i 0 
By equating the terms with identical powers of p one can have:
p 0 : u 0 ( x)  y0 ( x)  0 (1.10.a)
p1 : u1 ( x)  y0 ( x)  u 2 0 ( x)  0 (1.10.b)
p 2 : u 2 ( x)  2u 0 ( x)u1 ( x)  0 (1.10.c)
p 3 : u3 ( x)  2u 2 ( x)u 0 ( x)  u 21 ( x)  0 (1.10.e)
For simplicity, let u 0 ( x)  y 0 ( x) , then equation (1.10.a) is automatically satisfied.
Let y 0 ( x)  1 be the initial approximation of the differential equation (1.9), then
u 0 ( x)  1
By substituting u0 and y 0 into equation (1.10.b) one can have:
u1 ( x )  1
and this implies that:
- 10 -
u1 ( x)   x
By substituting u 0 and u1 into equation (1.10.c) one can have:
u 2 ( x)  2 x
u 2 ( x)  x 2
By substituting u1 , u 2 and u 3 into equation (1.10.e) one can have:
u 3 ( x)  3 x 3
u3 ( x)   x 3
By continuing in this manner one can have:
ui ( x)  (1) i x i , i = 0, 1, …
By substituting these functions into equation (1.8) one can obtain:


y ( x)   (1) i x i
i 0
1
 .
1 x
which is the exact solution of the ordinary differential equation (1.9).
Example (1.2.2):
y ( x)   y ( x)  x y ( x)  0 (1.11)
- 11 -
Here A( y )  y  ( y  x) y  and f ( x)  0. The operator can be divided into two parts L and
N, where L( y )  y and N ( y )  ( y  x) y 
u ( x )  y0 ( x )  py 0 ( x )  pu ( x )  x u ( x )  0 , p  [0, 1]
By substituting equation (1.7) into the above equation one can have:

  p i u ( x)  x    p i u ( x)  0
 p u ( x)  y  ( x)  py
i 0
i
i 0 0 ( x )  p

i 0
i
 
i 0
i

p 0 : u 0 ( x)  y0 ( x)  0 (1.12.a)
p1 : u1 ( x)  y0 ( x)  (u 0 ( x)  x)u 0 ( x)  0 (1.12.b)
p 2 : u 2 ( x )  u 0 ( x )  x u1 ( x )  u1 ( x )u 0 ( x )  0 (1.12.c)
Let y0 ( x)  x , then from equation (1.12.a) one can have:
u0 ( x)  x .
From equation (1.12.b) one can have:
u1 ( x)  3x
Therefore, the first approximation of equation (1.11) is:
y ( x)  u 0 ( x)  u1 ( x)  2 x
That is, the first approximation in this example is sufficient to give the exact solution. Next, if
we choose y0 ( x)  1, then u 0 ( x)  1, from equation (1.12.b) one can have:

- 12 -
u1 ( x)  1
and from equation (1.12.c) one can get:
u 2 ( x)  0
and in a similar manner one can obtain:
u i ( x )  0, i  3,4,...
so

y ( x)   ui ( x)  0
i 0
Example (1.2.3), [14]:
y ( x)  y 2 ( x)  0, 0  x  1 (1.13.a)
together with the initial condition
y(0)  1 (1.13.b)
Here A( y ( x ))  y ( x )  y ( x ) and f ( x)  0.
2
To solve this example by the homotopy perturbation method, consider equations (1.10):
For simplicity, let u 0 ( x)  y0 ( x)  y (0)  1, then equation (1.10.a) is automaticly
satisfied.
Since
- 13 -

y ( x)   ui ( x)
i 0
Then

y ( 0)   u i ( 0)
i 0
But u 0 (0)  1, therefore u i (0)  0, i  1,2,.... Thus, equation (1.10.b) becomes:
u1 ( x)   y0 ( x)  u 2 0 ( x)
 1.
By integrating both sides of the above ordinary differential equation from 0 to x and using the
initial condition u 1 (0)=0 one can get:
u1 ( x )   x .
From equation (1.10.c) one can have:
u 2 ( x )   2u 0 ( x )u1 ( x )
 2x.
Then by integrating both sides of the above differential equation from 0 to x and by using the
initial condition u 2 (0)=0, one can get:
u 2 ( x)  x 2 .
By continuing in this manner one can obtain:
ui ( x)  (x)i , i  0,1,....
Therefore
- 14 -

y ( x)   ui ( x)  1  x  x 2  
i 0
1 .

1 x
which is the exact solution of the above initial value problem.
Example (1.2.4):
y( x)  y( x) y( x)  1 (1.14.a)
together with the initial condition
y(0)  1 (1.14.b)
Here A( y )  y  yy  and f ( x)  1. The operator A can be divided into two parts L and N,
where L( y )  y and N ( y )  yy 
u ( x)  y 0 ( x)  py 0 ( x)  p[u ( x)u ( x)  1]  0, p  [0,1]
By substituting equation (1.7) into the above equation one can have:

 i 

 i p i
u ( x )  y 0 ( x )  py 0 ( x )  p   p u i ( x )  p i
u 
i ( x )  10
i 0 i  0 i 0 
p 0 : u0 ( x)  y0 ( x)  0 (1.15.a)
p1 : u1 ( x)  y0 ( x)  u 0 ( x)u 0 ( x)  1  0 (1.15.b)

- 15 -
Let y 0 ( x)  1 , then from equation (1.15.a) one can have:
u 0 ( x)  1
u1 ( x)  0
By continuing in this manner one can obtain:
ui ( x)  0, i  2,3,....
Therefore

y ( x)   ui ( x)  1.
i 0
which is the exact solution of the above initial value problem.
- 16 -
Chapter Two
Method for Solving the Integral
and Integro-Differential
Equations
Chapter Two The Homotopy Perturbation Method for Solving the Integral and Integro-Differential Equations
Introduction:
Many researchers used analytical methods and numerical methods for solving linear
and nonlinear integral and integro-differential equations [24], [26], [28].
The aim of this chapter is to use the homotopy perturbation method for solving special
types of linear and nonlinear integral and integro-differential equations of the second kind.
This chapter consists of three sections:
In section one, we describe the homotopy perturbation method for sovling the linear
Fredholm and Voltera integral equations of the second kind with its convergence.
In section two and three, we use the homotopy perturbation method to solve special
types of non-linear Fredholm integral and integro-differential equations of the second kind.
2.1 The Homotopy Perturbation Method for Solving Linear Integral Equations, [2]:
In this section, we use the homotopy perturbation method to solve the linear integral
equations of the second kind. To do this, first, consider the linear Fredholm integral equation
of the second kind:
b
u ( x)  f ( x)    k ( x, t )u (t )dt, x  [a, b] (2.1)
a
where f and k are known functions. The function f is said to be the driving term and k is said
to be the kernel function that depends on x, t and λ is a scalar parameter, a and b are known
constants and u is the unknown function that must be determined.
We rewrite equation (2.1) as:
- 17 -
A(u )  f ( x )  0 (2.2)
where A(u )  u ( x )    k ( x, t )u (t ) dt .
a
Then the integral operator A can be divided into two parts L and N such that equation (2.2)
becomes:
L(u )  N (u )  f ( x)  0 (2.3)
b
where Lu  u and N    k ( x, t )udt.
a
Accoding to [13], we construct a homotopy v : [a, b]  [0,1] 

  which satisfies:
 b

H (v, p)  (1  p)v( x, p)  u0 ( x)  p v( x, p)    k ( x, t )v(t , p)dt  f ( x)  0 (2.4)
 a 
where p [0,1],  represents the set of all real numbers and u 0 is the initial approximation
to the solution of equation (2.1).
By using equation (2.4) it follows that:
H ( v ,0 )  v ( x ,0 )  u 0 ( x )  0
b
H (v,1)  v ( x,1)    k ( x, t )v (t ,1) dt  f ( x )  0
a
and the changing process of p from zero to unity is just that of v( x, p ) from u 0 ( x) to u (x ).
Therefore
b
v( x,0)  u 0 ( x)  v( x,1)    k ( x, t )v(t ,1)dt  f ( x), x  [ a, b]
a
- 18 -
and
u 0 ( x)  u ( x), x  [a, b].
Next, we assume that the solution of equation (2.4) can be expressed as

v ( x, p )   p i v i ( x ) (2.5)
i 0
Therefore the approximated solution of the integral equation (2.1) can be obtained as follows:
u ( x )  lim v ( x , p )
p 1
 (2.6)
  vi ( x )
i 0
By substituting the approximated solution given by equation (2.5) into equation (2.4) one can
get:

 b 


i 0
p i
v i
( x )  u 0
( x )  pu 0
( x )  p 
 a  k ( x , t ) 
i 0
p i
v i
(t )dt  f ( x )   0
Then by equating the terms with identical powers of p one can have:
p 0 : v0 ( x )  u 0 ( x )  0 (2.7.a)
b
p : v1 ( x)  u0 ( x)  f ( x)    k ( x, t )v0 (t )dt  0
1
(2.7.b)
a
b
p : v j ( x)    k ( x, t )v j 1 (t )dt  0, j  2,3,...
j
(2.7.c)
a
For simplicity we set v0 ( x )  u 0 ( x )  f ( x ), then equation (2.7.a) is automatically satisfied.
By substituting u 0 ( x )  v0 ( x )  f ( x ) into equation (2.7.b) one can have:
- 19 -
b
v1 ( x )    k ( x, t ) f (t )dt
a
By substituting v 1 into equation (2.7.c) one can have:
b
v 2 ( x )    k ( x, t )v1 (t )dt
a
In a similar manner one can get vi ( x), i  3,4,.... By substituting vi ( x ), i  0,1,... into
equation (2.6) one can get the approximated solution of the integral equation (2.1).
Next, we study the convergence of the homotopy perturbation method for solving the
integral equation (2.1). To do this, consider the iteration formula that is obtained by applying
the homotopy perturbation method to solve the integral equation (2.1):
b
vi ( x)    k ( x, t )vi 1 (t )dt , i  1,2,... (2.8)
a
with the initial approximation v0 ( x )  f ( x )
According to the previous equation, we define the partial sum as follow:
n
sn ( x)   vi ( x), n  0,1,... (2.9)
i 0
where s0 ( x )  f ( x ).
In view of equations (2.8) and (2.9), one can have:
s0 ( x )  f ( x )
and
- 20 -
n 1 n
s n1 ( x)   vi ( x)  v0 ( x)   vi ( x)
i 0 i 0
b b b
 f ( x)    k ( x, t )v0 (t )dt    k ( x, t )v1 (t )dt      k ( x, t )vn (t )dt
a a a
b
 f ( x)    k ( x, t )v0 (t )  v1 (t )    vn (t )dt
a
b
 f ( x)    k ( x, t ) s n (t )dt.
a

From [21], it is known that if v1 , v2 ,... be a sequence of functions, then the series  vi ( x ) is
i 1
said to be convergence to u if the sequence s n  of partial sums defined by:
n
s n ( x )   vi ( x )
i 0
converges to u.
Now, we are in the position that we can give the following theorem.
Theorem (2.1), [21]:
Consider the iteration scheme:
s0 ( x )  f ( x )
and
b
sn1 ( x)  f ( x)    k ( x, t ) sn (t )dt , n  0,1,...
a
- 21 -
to construct a sequence of successive iterations sn (x) to the solution of equation (2.1). Let
f  L2 (a, b) and
bb
  k ( x, t )
2
dxdt  B 2  .
aa
1
If   , then the above iteration scheme convergence to the solution of equation (2.1).
B
To illustrate this method, consider the following example.
Example (2.2):
Consider the following linear Fredholm integral equation of the second kind:
1
1
u ( x)  e3 x  (2 e3  1) x   xtu(t )dt, 0  x 1
9 0
Here a  0, b    1, f(x)  e 3x 
x 3
9
 
2e  1 and k ( x, t )  xt.
Therefore
bb 11
1
  k ( x, t ) dxdt    x 2t 2 dxdt 
2
 B 2  .
aa 00 9
and
1
 1  3.
B
So, we can use the homotopy perturbation method to solve this example.
- 22 -
To do this, let
u 0 ( x )  v0 ( x )  f ( x )  e 3 x 
x 3
9

2e  1 . 
Hence
b
v1 ( x, t )    k ( x, t )v0 (t )dt
a

 

1
t
  xt e 3t  2e 3  1 dt
0  9 
 2  4e 3 
  x.
 27 
and
b
v2 ( x, t )    k ( x, t )v1 (t )dt
a
1
 2  4e 3 
  xt 2  dt
0  27 
 2  4e 3 
  x.
 81 
By continuing in this manner, one can have:
b
vi ( x, t )    k ( x, t )vi 1 (t )dt
a
 2  4e 3 
 i 1 
x, i  1,2,....
 27 (3) 
Thus
- 23 -

u ( x )   vi ( x )
i 0
 
x 3   2  4e 3 
e 3x
 2e  1    i 1 
x
9 i 1  27(3) 
 2  4e 3    1  i 1
e 3x x 3

 2e  1     x  
9  27  i 1  3 
 e 3x
x 3

 2e  1  
9

3  2  4e 3 
2  27 
x
 e 3x .
which is the exact solution of the above integral equation.
Example (2.3):
Consider the following linear Fredholm integral equation of the second kind:
1
8 3 4
u ( x)  e 3x
 e x  x  4 xtu(t )dt, 0  x 1
9 9 0
8 3 4
Here a  0, b  1,   4, f(x)  e 3x  e x  x and k ( x, t )  xt.
9 9
Therefore
bb 11
1
  k ( x, t ) dxdt    x 2t 2 dxdt 
2
 B 2  .
aa 00 9
and
1
 4  3.
B
- 24 -
So, if we use the homotopy perturbation method to solve this example, then the solution that
is obtained by applying this method may be convergent to the exact solution or may not. To
do this, let
8 4
u 0 ( x )  v0 ( x )  f ( x )  e 3 x   e 3   x
9 9
Hence
b
v1 ( x, t )  4  xtf (t )dt
a
 4 
1
8
 4  xt e 3t  e 3t  t  dt
0  9 9 
 8 4
  e3   x
 27 27 
and
b
v2 ( x, t )  4  xtv1 (t )dt
a
 8 4
1
 4  xt 2  e 3  dt
0  27 27 
  32 3 6 
 e   x.
 81 81
1
vi ( x, t )  4  xtvi 1 (t )dt
0
  8(4) i 1 3 4(4) i 1 
 i 1
e  i 1 
x, i  1,2,....
 27 (3) 27 (3) 
Thus
- 25 -

u ( x )   vi ( x )
i 0
   8( 4) i 1 4(4) i 1 
8 4
 e 3x   e 3   x    i 1
e 3
 i 1 
x.
 9 9  i 1  27 (3) 27 (3) 
Since
  8  4  i 1 3 4  4  i 1    8 3 4    4  i 1
  27  3  e  27  3  x   27 e  27  x  3  .
  
i 1       i 1  
 i 1 
4 4
But    is a geometric series, that is divergent since r   1. Therefore  vi ( x) is
i 1  3  3 i 0
divergent.
Next, consider the linear Volterra integral equation of the second kind:
x
u ( x)  f ( x)    k ( x, t )u (t )dt (2.10)
a
where k is the kernel of the integral equation, f is the driving term,  is a scalar parameter, a
is a known constant and u is the unknown function that must be determined.
To solve this integral equation via the homotopy perturbation method, we rewrite
equation (2.10) as:
A(u )  f ( x )  0 (2.11)
x
where A(u )  u ( x)    k ( x, t )u (t )dt .
a
Then A can be divided into two parts L and N such that equation (2.11) becomes:
- 26 -
L(u )  N (u )  f ( x)  0
x
where Lu  u and Nu    k ( x, t )u (t ) dt .
a
Accoding to [13], we construct a homotopy v : [a, )  [0,1] 

 x

H (v, p)  (1  p)v( x, p)  u0 ( x)  pv( x, p)    k ( x, t )v(t , p)dt  0 (2.12)
 a 
where p [0,1]  represents the set of all real numbers and u 0 is the initial approximation to
the solution of equation (2.10).
H ( v ,0 )  v ( x ,0 )  u 0 ( x )  0
x
H (v,1)  v ( x,1)    k ( x, t )v (t ,1) dt  f ( x )  0
a
and the changing process of p from zero to unity is just that of v( x, p ) from v( x,0)  u ( x) to
x
v ( x,1)    k ( x, t )v (t ,1) dt  f ( x ).
a
Therefore
x
v( x,0)  u0 ( x)  v( x,1)    k ( x, t )v(t ,1)dt  f ( x), xa
a
and
u 0 ( x)  u ( x), x  a.
- 27 -
Next, we assume that the solution of equation (2.12) can be expressed as in equation
(2.5). By substituting this approximated solution into equation (2.12) one can get:
  x  
 i p v i
( x )  u 0 ( x )  pu 0 ( x )  p  
  k ( x , t )  i p i
v (t )dt  f ( x ) 0
i 0  a i  0 
p 0 : v0 ( x )  u 0 ( x )  0
x
p : v1 ( x)  u0 ( x)  f ( x)    k ( x, t )v0 (t )dt  0
1
x
p : v j ( x)    k ( x, t )v j 1 (t )dt  0, j  2,3,...
j
So, if we choose v0 ( x )  u 0 ( x )  f ( x ), then one can get the iteration formula.
v0 ( x )  f ( x )
x
vi ( x)    k ( x, t )vi 1 (t )dt , i  1,2,...
a
Next, we study the convergence of the homotopy perturbation method for solving the
integral equation (2.10). To do this, consider the following theorem.
Theorem (2.4), [21]:
Consider the iteration scheme:
s0 ( x )  f ( x )
and
- 28 -
x
sn1 ( x)  f ( x)    k ( x, t ) sn (t )dt , n  0,1,...
a
to construct a sequence of successive iterations s n (x ) to the solution of the integral equation
(2.10).
If f and k are real-valued continuous functions, then the above iteration scheme convergence
to the solution of the integral equation (2.10) for all values of  .
To illustrate this method consider the following example.
Example (2.5):
Consider the following linear Volterra integral equation of the second kind:
x
u ( x)  x    ( x  t )u (t )dt
0
To solve this example via the homotopy perturbation method, consider the iteration
formula:
v0 ( x )  u 0 ( x )  x.
x
vi 1 ( x)    ( x  t )vi (t )dt , i  0,1,....
0
Therefore
x

v1 ( x)    t ( x  t )dt  x3
0 3!
- 29 -
and
2 x
v2 ( x)  t ( x  t )dt
3
3! 0
 2
 x5.
5!
i
vi ( x , t )  x 2i 1 , i  1,2,....
(2i  1)!
It is known that the sequence
n x 2i 1
sn ( x)   i
i 0 (2i  1)!
which is convergent for all values of  and x .
Therefore
 x 2i 1
u ( x)    i
.
i 0 (2i  1)!
which is the exact solution of the above integral equation.
Note that, if   1, then u ( x)  sin( x).
2.2 The Homotopy Perturbation Method For Solving Non-Linear Integral Equations:
In this section, the homotopy perturbation method is used to solve special types of non-
linear Fredholm integral equations with some illustrative examples. To do this, consider the
following non-linear Fredholm integral equation of the second kind:
- 30 -
b
u ( x)  f ( x)    k ( x, t )u (t ) dt , axb
q
(2.13)
a
where q  , f is the driving term and k is the kernel of the integral equation that depends on
x and t, λ is a scalar parameter, a and b are known constants and u is the unknown function
that must be determined.
We rewrite equation (2.13) as:
A(u )  f ( x)  0 (2.14)
b
where A(u )  u ( x)    k ( x, t )u (t ) dt.
q
Then the integral operator A can be divided into two parts L and N equation (2.14)
becomes:
L(u )  N (u )  f ( x)  0
b
where L (u )  u and N (u )     k ( x , t )u (t )  tdt .
q
Accoding to [13], we can construct a homotopy v : a , b   0,1 

 b

H (v, p )  (1  p )v( x, p )  u 0 ( x)  p v( x, p )    k ( x, t )v(t , p ) dt  f ( x)  0
q
(2.15)
 a 
where p [0,1],  represents the set of all real numbers and u 0 is the initial approximation
to the solution of equation (2.13).
H ( v ,0 )  v ( x ,0 )  u 0 ( x )  0
- 31 -
b
H (v,1)  v( x,1)    k ( x, t )v(t ,1) dt  f ( x)  0
p
and the changing process of p from zero to unity is just that of v(v, p) from u0 ( x) to u(x).
Therefore
b
v( x,0)  u0 ( x)  v( x,1)    k ( x, t )v(t ,1) dt  f ( x), x  [ a, b]
q
and
u 0 ( x)  u ( x), x  [a, b].
Next, we assume that the solution of equation (2.15) can be expressed as in equation
(2.5). By substituting the approximated solution given by equation (2.5) into equation (2.15)
one can get:
  b  i 
q

 p vi ( x)  u0 ( x)  pu0 ( x)  p    k ( x, t )  p vi (t ) dt  f ( x)  0
i
i 0  a  i 0  
p 0 : v0 ( x )  u ( x )  0 (2.16.a)
b
p : v1 ( x)  u 0 ( x)  f ( x)    k ( x, t )v0 (t ) dt  0
1 q
(2.16.b)
a
- 32 -
 b
 v2 ( x)    k ( x, t )2v0 (t )v1 (t )dt  0, if q  2
 a

 
b
 v ( x)   k ( x, t ) 3(v (t )) 2 v (t ) dt  0, if q  3
p2 :  2  0 1
(2.16.c)
a

 
b
 v ( x)   k ( x, t ) 4(v (t )) 3 v (t ) dt  0, if q  4
 2  0 1
 a


 
b j 1
 v j ( x)    k ( x, t )  vk (t )v j k 1 (t ) dt  0, if q  2
 a k 0

 
b j 1 j i 1
 v ( x)   k ( x, t )   v (t )v (t )v
p j :
j  i k j  k i 1 (t ) dt  0, if q  3
(2.16.e)
a i  0 k  0

 
b j 1 j i 1 j i  k 1
 v ( x )   k ( x, t )
 j     vi (t )vk (t )vl (t )v j l k i1 (t ) dt  0, if q  4
i 0 k 0 l 0
 a

where j=3,4,….
For simplicity, we set v 0 ( x)  u 0 ( x)  f ( x) , then equation (2.16.a) is automatically
satisfied. By substituting v 0 ( x)  u 0 ( x)  f ( x) into equation (2.16.b) one can have:
b
v1 ( x)    k ( x, t )v0 (t ) dt  0
q
By substituting v0 ,v1 into equation (2.16.c) one can get v2 ( x).
In a similar manner, one can get vi ( x ), i  3,4,... By substituting vi ( x ), i  0,1,... into equation
(2.6) one can get the approximated solution of the integral equation (2.13).
To illustrate this method consider the following examples.
- 33 -
Example (2.4):
Consider the following nonlinear Fredholm integral equation of the second kind:
1 1
u ( x)  x 2    ( x 2  t )u (t ) dt, 0  x  1
4 2
(2.17)
5 6 0
4 2 1
Here a  0, b    1, q  2, f(x)  x  and k ( x, t )  x 2  t.
5 6
To solve this example via the homotopy perturbation method, let
4 2 1
v0 ( x )  u 0 ( x )  f ( x )  x 
5 6
 0.8x 2  0.16667.
Then
b
v1 ( x)   k ( x, t )v0 (t )  dt
q
a
2
4 1
1
  ( x  t )  t 2   dt
2
0 5 6
 0.18722  0.24467 x 2 .
In this case, let N=1, then

u ( x)   vi ( x)  v0 ( x)  v1 ( x)  1.04466 x 2  2.05555  10 2
i 0
Next, we must find v2 ( x ) :
b
v2 ( x)   k ( x, t )2v0 (t )v1 (t )dt
a
1
4 1    337 367 2 
 2 ( x 2  t )  t 2     t  dt
0  5 6  1800 1500 
 0.02046  0.05678 x 2 .
- 34 -

u ( x)   vi ( x)  v0 ( x)  v1 ( x)  v2 ( x)
i 0
 0.98789 x 2  2.62963  105
The following table gives the approximated solution of example (2.4) for different values of
N.
Table (1.1) represents the approximated solutions of example (2.4) for different values of N.
N u (x) N u(x) N u(x)
0 0.8x2+0.166667 4 1.00222x2–1.26717×10-3 8 0.99993x2+1.31635×10-5
1 1.04466x2–2.05555×10-2 5 0.99924x2+6.40980×10-5 9 1.00001x2+1.52175×10-5
2 0.98789x2–9.62963×10-5 6 0.99994x2+2.05376×10-4 10 1.00001x2–1.03366×10-5
3 0.99762x2+3.58386×10-3 7 1.00016x2–1.06638×10-4 11 1.00001x2–1.00453×10-5
Note that from the above table one can deduce that as N increases, the approximated solution
of the integral equation (2.17) converges to the exact solution u ( x)  x 2 .
2.3 The Homotopy Perturbation Method for Solving Non-Linear Fredholm Integro-
Differential Equations, [26]:
In this section, the homotopy perturbation method is employed for solving the initial value
problems of special types of the first order non-linear Fredholm integro-differential equations
with some illustrative examples.
- 35 -
To do this, consider the following first order nonlinear Fredholm integro-differential
equation of the second kind:
b
u ( x)  f ( x)    k ( x, t )u (t ) dt , axb
q
(2.18.a)
a
together with the initial condition:
u (a )   (2.18.b)
where q  , f is the driving term and k is the kernel of the integro-differential equation that
depends on x and t, λ is a scalar parameter, a and b are known constants and u is the
unknown function that must be determined.
It is clear that, if q=1, then the integro-differential equation is linear, otherwise it is
nonlinear.
We rewrite equation (2.18.a) as:
A(u )  f ( x)  0 (2.19)
b
   k ( x, t )u (t ) dt .
du
where A(u )  q
dx a
Then the operator A can be divided into two parts L and N such that equation (2.19)
becomes:
L(u )  N (u )  f ( x)  0
b
and N (u )    k ( x, t )u (t )q dt .
du
where L(u ) 
dx a
According to [13], we construct a homotopy v : a , b   0,1 

- 36 -
 dv( x, p ) du 0 ( x)   dv( x, p ) b

H (v, p )  (1  p )       k ( x, t )v(t , p ) dt  f ( x)  0 (2.20)
q
p
 dx dx   dx a 
where p [0,1]  represents the set of all real numbers and u 0 is the initial approximation to
the solution of equation (2.18.a) which satisfies the initial condition given by equation
(2.18.b).
dv( x,0) du 0 ( x)
H ( v ,0 )   0
dx dx
b
   k ( x, t )v(t ,1) dt  f ( x)  0
dv( x,1)
H (v,1) 
q
dx a
and the changing process of p from zero to unity is just that of v( x, p ) from u 0 ( x) to u (x).
Therefore
v( x,0) v0 ( x) v( x,1) b

     k ( x, t )v(t ,1) dt  f ( x), x  [ a, b]
q
x x x a
and
u 0 ( x)  u ( x), x  [a, b].
Next, we assume that the solution of equation (2.18.a) can be expressed as in equation
(2.5).
By substituting the approximated solution given by equation (2.5) into equation (2.20) one
can get:
 dvi ( x) du 0 ( x) du ( x)  b  i 
q

 p dx i
 p 0  p    k ( x, t )  p vi (t ) dt  f ( x)  0
i 0 dx dx  a  i 0  
- 37 -
dv0 ( x) du 0 ( x)
p0 :  0 (2.21.a)
dx dx
b
 f ( x)    k ( x, t )v0 (t ) dt  0
dv1 ( x) du0 ( x)

1 q
p : (2.21.b)
dx dx a
 dv2 ( x) b
    k ( x, t )2v0 (t )v1 (t )dt  0, if q  2
 dx a

 
b
 dv2 ( x)   k ( x, t ) 3(v (t )) 2 v (t ) dt  0, if q  3
p :  dx
2  0 1
(2.21.c)
a

 
b
 dv2 ( x)   k ( x, t ) 4(v (t )) 3 v (t ) dt  0, if q  4
 dx  0 1
 a

 dv j ( x)
 
b j 1
    k ( x, t )  vk (t )v j k 1 (t ) dt  0, if q  2
 dx a i 0
 dv ( x)
 
b j 1 j i 1
 j    vi (t )vk (t )v j k i1 (t ) dt  0,
p j :  dx
  k ( x , t ) if q  3
(2.21.e)
a i 0 k 0

 
b j 1 j i 1 j i  k 1
 dv j ( x)   k ( x, t )
 dx     vi (t )vk (t )vl (t )v j l k i1 (t ) dt  0, if q  4
i 0 k 0 l 0
 a

where j=3,4,….
x
Since u (a )   then we choose u 0 ( x)     f (t )dt and this implies that u 0 (a )   . Also, for
a
simplicity we set v0 ( x)  u 0 ( x)     f (t )dt. So equation (2.21.a) is automatically satisfied.

a
Therefore by substituting x  a in equation (2.6) one can have:
- 38 -

u ( a )   vi ( a )
i 0
But v0 (a )  u (a )   , hence vi (a )  0, i  1,2,.... By substituting v0 ( x)  u 0 ( x)     f (t )dt

a
into equation (2.21.b) one can get:
q
dv1 ( x) b
 t

   k ( x, t )    f ( z )dz  dt  0
dx a  a 
By integrating both sides of the above differential equation and by using the initial condition
u1 (a )  0 one can obtain:
q
xb
 t

v1 ( x)     k ( s, t )    f ( z )dz  dtd s  0 (2.22)
aa  a 
By substituting v0 and v1 into equation (2.21.c) and by solving the resulting first order linear
ordinary differential equation together with the initial condition v2 (a )  0 one can get v2 ( x) .
Then by substituting j= 3 v0 , v1 and v2 into equation (2.21.e) and by using initial condition
v3 (a )  0 one can solve the resulting first order linear ordinary differential equation to get
v3 ( x) . In a similar manner one can get vi ( x ), i  4,5,.... By substituting vi ( x ), i  0,1,... into
equation (2.6) one can get the approximated solution of the initial value problem given by
equations (2.18).
To illustrate this method consider the following examples.
- 39 -
Example (2.5), [28]:
Consider the initial value problem that consists of the first order linear Fredholm
integro-differential equation of the second kind:
1 5 1
u ( x )   x   ( xt  1)u (t ) dt , 0  x 1 (2.23.a)
2 6 0
u ( 0)  1 (2.23.b)
1 5
Here a  0, b    1, f(x)   x and k ( x, t )  xt  1.
2 6
We use the homotopy perturbation method to solve this example. To do this, let
1 5 
b x
v0 ( x)  u 0 ( x)     f (t )dt  1     t  dt
a 0 2 6 
1 5
1 x  x2
2 12
 1  0.5x  0.41667x 2 .
Then
 1 5 
x1
v1 ( x)    ( st  1) 1  t  t 2 dtds
00  2 12 
 0.11458 x 2  1.375 x.
N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)  1  0.11111x  0.30208 x 2
i 0
- 40 -
Next, we must find v2 ( x)
 
x1
v2 ( x)    ( st  1) 0.11458x 2  1.375 x dtds
00
 0.11617x 2  0.34375x.
Therefore, for N=2,
N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)  v2 ( x)
i 0
 1  0.45486x  0.18591x 2
In a similar manner one can get vi ( x), i  3,4,... .The following table gives the approximated
solutions for different values of N.
N u(x) N u(x) N U(X)
0 1–0.5x–0.41667x2 9 1+0.98903x–3.7401×10-3x2 18 1+0.99986x–4.6173×10-5x2
1 1+0.11111x–0.30208x2 10 1+0.99327x–2.2952×10-3x2 19 1+0.99992x–2.8335×10-5x2
2 1+0.45486x–0.18591x2 11 1+0.99587x–1.4085×10-3x2 20 1+0.99995x–1.7389×10-5x2
3 1+0.79470x–0.070018×10-2x2 12 1+0.99747x–8.6440×10-4x2 21 1+0.99997x–1.0671×10-5x2
4 1+0.87401x–4.2969×10-2x2 13 1+0.99844x–5.3047×10-4x2 22 1+0.99998x–6.5488×10-6x2
5 1+0.92268x–2.6369×10-2x2 14 1+0.99905x–3.2554×10-4x2 23 1+0.99999x–4.0189×10-6x2
6 1+0.95255x–1.6183×10-2x2 15 1+0.99941x–1.9978×10-4x2 24 1+0.99999x–2.4663×10-6x2
7 1+0.97088x–9.9310×10-3x2 16 1+0.99964x–1.2260×10-4x2 25 1+x–1.5136×10-6x2
8 1+0.98213x–6.0945×10-3x2 17 1+0.99978x–7.5238×10-5x2 26 1+x–1.00336×10-6x2
- 41 -
Note that from the above table one can deduce that as N increases the approximated solution
of the initial value problem given by equations (2.23) converges to the exact solution
u(x)=1+x.
Example (2.6), [26]:
Consider the initial value problem that consider of the first order nonlinear Fredholm integro-
differential equation:
1
x   ( x 2  t )u (t ) dt
159 1 2 2
u ( x)  
3
(2.24.a)
160 64 0
u (0)  0 (2.24.b)
1 159 1 2
Here a=0, b= , q=3,   1 , f ( x)   x and k ( x, t )  x 2  t .
2 160 64
156 1 2 
x
v0 ( x )  u 0 ( x )     t  dt
0 160 64 
159 1 3
 x x .
160 192
Then
1
3
159 1 3
x2
v1 ( x)    ( s  t ) 
2
t t  dtds .
00 160 192 
 6.11634  10 3 x  5.09791  10 3 x 3 .
- 42 -

N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)
i 0
 0.99987 x  1.10428  10 4 x 3 .
Next, we must find v2 ( x)
1
x2
v2 ( x)    ( s 2  t )3v0 (t )v1 (t )dtds
00
 8.54066  10 4 x  8.38354  10 4 x 3 .
Therefore, for N=2,

N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)  v2 ( x)
i 0
 1.00072x - 7.27926 × 10 -4 x 3 .
In a similar manner one can get vi ( x), i  3,4,... . The following table gives the approximated
solutions of example (2.6) for different values of N.
N u(x) N u(x) N u(x)
0 0.99375x-5.20833×10-3x3 4 1.00086x+8.63897×10-4x3 8 1.00086x+8.66701×10-4x3
1 0.99987x-1.10428×10-4x3 5 1.00086x+8.66308×10-4x3 9 1.00086x+8.66702×10-4x3
2 1.00072x-7.27926×10-4x3 6 1.00086x+8.66651×10-4x3 10 1.00086x+8.66703×10-4x3
3 1.00084x+8.46953×10-4x3 7 1.00086x+8.66700×10-4x3 11 1.00086x+8.66704×10-4x3
- 43 -
Note that from the above table one can deduce that as N increases the approximated
solution of the initial value problem given by equations (2.24) converges to the exact solution
u(x)=x.
Example (2.7), [26]:
Consider the initial value problem that consists of the nonlinear first order Fredholm
integral differential equation:
1 1
x    ( xt  1)u (t ) dt
11
u ( x) 
2
(2.25.a)
6 5 0
u (0)  0 (2.25.b)
11 1
Here a=0, b=  =1, q=2, f ( x)  x  and k ( x, t )  xt  1 .
6 5
x
11 1
v0 ( x)  u 0 ( x)   ( t  )dt
0 6 5
11 2 1
 x  x.
12 5
Then
2
11 1 
x1
v1 ( x)    ( s t  1)  t 2  t  dtds
2
00 12 5 
 0.03836 x 2  0.8975 x. .
- 44 -

N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)
i 0
 0.95502 x 2  0.11028 x .
Next, we must find u 2 ( x )
x1
v2 ( x)    k ( s, t )2v0 (t )v1 (t )dtds
00
 0.01629 x 2  0.03939 x. .
Therefore, for N=2,

N
u ( x)   vi ( x)  v0 ( x)  v1 ( x)  v2 ( x)
i 0
 0.97131x 2  7.09997 × 10 -2 x.
In a similar manner one can get vi ( x ), i  3,4,... . The following table gives the approximated
solutions for different values of N.
- 45 -
Table (2.4) represents the approximated solutions of example (2.7) for different values of N
N U(x) N u(x) N u(x)
0 0.91667x2–0.2x 13 0.998330x2–4.16697×10-3x 26 0.999838x2–4.04670×10-4x
1 0.95502x2–0.11028x 14 0.998631x2–3.41862×10-3x 27 0.999863x2–3.43486×10-4x
2 0.97131x2–7.09997×10-2x 15 0.998872x2–2.81678×10-3x 28 0.999883x2–2.91965×10-4x
3 0.98022x2–4.90465×10-2x 16 0.999067x2–2.32982×10-3x 29 0.999901x2–2.48517×10-4x
4 0.98572x2–3.54806×10-2x 17 0.999226x2–1.93369×10-3x 30 0.999915x2–2.11818×10-4x
5 0.98936x2–2.64527×10-2x 18 0.999355x2–1.60990×10-3x 31 0.999928x2–1.80774×10-4x
6 0.99190x2–2.01584×10-2x 19 0.999462x2–1.34412×10-3x 32 0.999938x2–1.54475×10-4x
7 0.99373x2–1.56202×10-2x 20 0.999550x2–1.12510×10-3x 33 0.999947x2–1.32165×10-4x
8 0.99508x2–1.22643×10-2x 21 0.999622x2–9.43993×10-4x 34 0.999954x2–1.13214×10-4x
9 0.996096x2–9.73292×10-3x 22 0.999682x2–7.93753×10-4x 35 0.999961x2–9.70962×10-5x
10 0.996875x2–7.79290×10-3x 23 0.999732x2–6.68758×10-4x 36 0.999966x2–8.33727×10-5x
11 0.997480x2–6.28651×10-3x 24 0.999774x2–5.64490×10-4x 37 0.999971x2–7.16747×10-5x
12 0.997955x2–5.10395×10-3x 25 0.999809x2–4.77297×10-4x 38 0.999975x2–6.16927×10-5x
Note that from the above table one can deduce that as N increases, the approximated
solution of the initial value problem given by equations (2.25) converges to the exact solution
u ( x)  x 2 .
- 46 -
Chapter Three
Method for Solving Some
Nonlocal Problems
Chapter Three The Homotopy Perturbation Method for Solving Some Nonlocal Problem
Introduction:
It is seen that in the modeling of many real life applications systems in various fields of
physics, ecology, biology, etc, an integral term over the spatial domain is appeared in some part
or in the whole boundary, [8]. Such boundary value problems are known as nonlocal problems.
The integral term may appear in the boundary conditions. Nonlocal conditions appear when
values of the function on the boundary are connected to values inside the domain, [3]
Many researchers studied the nonlocal problems, say, [7] used Galerkin method for
solving the nonlocal problem for the diffusion equation, [6] discussed the existence of the
solutions for the nonlocal problem of the one-dimensional wave equations, [30] used Fourier
method to establish the existence of the solution for a class of linear hyperbolic equations with
nonlocal conditions, [27] used the homotopy perturbation method for solving the one-
dimensional parabolic integro-differential equations with some real life applications.
In this chapter, we use the homotopy perturbation method to solve some types of the
nonlocal problems.
This chapter consists of two sections:
In section one, we use the homotopy perturbation method for solving the one-dimensional
wave equation with non-homogeneous Neumann and nonlocal conditions.
In section two, we give the solution of hyperbolic integro-differential equations with non-
homogeneous Neumann and nonlocal conditions via the homotopy perturbation method.
- 47 -
3.1 Solutions of One-Dimensional Wave Equation with Non-Homogeneous Neumann
and Nonlocal Conditions:
Consider the one-dimensional non-homogeneous wave equation:
 2 u ( x, t )  2 u ( x, t )
  f ( x, t ), (x, t)   (3.1.a)
t 2 x 2
together with the initial conditions:
u ( x ,0)  r1 ( x ), 0  x  , (3.1.b)
u ( x, t )
 r2 ( x), 0  x  , (3.1.c)
t t 0
the non-homogeneous Neumann condition:
u ( x, t )
  (t ), 0  t  T (3.1.d)
x x 0
and the non-homogeneous nonlocal condition:
 u ( x, t )dx   (t ), 0t T (3.1.e)

0
where f is a known function of x and t,   ( x, t ) 0  x  , 0  t  T , r1 , r2 , and  are given
functions that must satisfy the compatibility conditions:
 
r1(0)   (0), r2 (0)   (0),  r1 ( x)dx   (0), and  r2 ( x)dx   (0).
0 0
To solve this nonlocal problem by the homotopy perturbation method, we first transform this
nonlocal problem into another nonlocal problem, but with homogeneous Neumann condition and
homogeneous nonlocal conditions. To do this we use the transformation that appeared in [1]:
- 48 -
w( x, t )  u ( x, t )  z ( x, t ), (x, t)   (3.2)
    (t )
where z ( x, t )   (t )  x    .
 2 
Then
 2 u ( x , t )  2 w( x , t )  2 z ( x , t )
 
t 2 t 2 t 2
and
 2 u ( x , t )  2 w( x , t )
 .
x 2 x 2
Therefore the nonlocal problem given by equations (3.1) is transformed to the one-dimensional
non-homogeneous wave equation:
 2 w( x, t )  2 w( x, t )
  g ( x, t ), (x, t)   (3.3.a)
t 2 x 2
w( x,0)  q1 ( x ), 0  x   (3.3.b)
w ( x , t )
 q 2 ( x ), 0  x   (3.3.c)
t t  0
the homogeneous Neumann condition:
w( x, t )
 0, t0 (3.3.c)
x x 0
and the homogeneous nonlocal condition:
 w( x, t )dx  0, t0 (3.3.d)

0
- 49 -
 z(x,t)
2
z ( x, t )
where g ( x , t )  f ( x , t )  , q1 ( x)  r1 ( x)  z ( x,0) and q 2 ( x)  r2 ( x)  .
t 2 t t 0
To solve this nonlocal problem by the homotopy perturbation method, we construct a
homotopy v :   [0,1] 
 which satisfies:
 2 v( x, t , p)  2 w0 ( x, t )  2 w0 ( x, t )   2 v( x, t , p) 
H (v, p)   p  p   g ( x, t )  0 (3.4)
t 2 t 2 t 2  x 2

where p  [0,1] ,  represents the set of all real numbers and w 0 is the initial approximation to
the solution of equation (3.3.a) which satisfies the initial conditions, the Neumann condition and
the nonlocal condition given by equations (3.3.b)-(3.3.d).
 2 v ( x , t ,0 )  2 w0 ( x , t )
H ( v ,0 )   0
t 2 t 2
 2 v ( x , t ,1)  2 v ( x , t ,1)
H ( v ,1)    g ( x, t )  0
t 2 x 2
Next, we assume that the solution of equation (3.4) can be expressed as:

w( x, t , p)   p i vi ( x, t ) (3.5)
i 0
Therefore the approximated solution of the nonlocal problem given by equations (3.3) can be
obtained as follows:

w( x, t )  lim v( x, t , p)   vi ( x, t ) (3.6)
p1 i 0
can get:
- 50 -
 vi ( x, t )  w0 ( x, t )
 2
 w0 ( x, t )
2
  i  2 vi ( x, t )
2

H (v, p)   p i
 p  p   p  g ( x, t )  0.
i 0 t 2
t 2
t 2
 i 0 x 2

 2 v0 ( x, t )  2 w0 ( x, t )
p : 0
 0 (3.7.a)
t 2 t 2
 2 v1 ( x, t )  2 w0 ( x, t )  2 v0 ( x, t )
p : 1
   g ( x, t )  0 (3.7.b)
t 2 t 2 x 2
 2 v j ( x, t )  2 v j 1 ( x, t )
p : j
  0, j  2,3,... (3.7.c)
t 2 x 2
For simplicity, we take v0 ( x, t )  w0 ( x, t ). In this case equation (3.7.a) is automatically satisfied.
Let w0 ( x, t )  q1 ( x)  q2 ( x)t then
w 0 ( x , 0 )  q1 ( x ), 0  x  ,
w0 ( x, t )
 q 2 ( x ), 0  x  ,
t t 0
 
w0 ( x, t ) z ( x , t )   2 z ( x, t ) 
 q1 (0)  q 2 (0)t  r1(0)    r2 (0)  t
x x0 x t  0 t x t  0 
x0  x0 
 
 r1(0)   (0)  r2 (0)t   (0)t  0, 0  t  T
and
    
 z ( x, t ) 
 0
w ( x, t ) dx   1
q ( x ) dx  t  2
q ( x ) dx   1
( r ( x )  z ( x,0))dx  t  r2 ( x)  t  dx
0 0 0 0 0 t 0 
  (0)   (0)   (0)t   (0)t  0, t  0.
Therefore w o satisfies the initial conditions, the Neumann condition and the nonlocal condition
given by equations (3.3.b)-(3.3.d). Therefore by substituting t=0 in equation (3.6) one can have:
- 51 -

w( x,0)   vi ( x,0).
i 0
But v0 ( x,0)  q1 ( x) and w( x ,0)  q1 ( x ), hence vi ( x,0)  0, i  1,2,.... By substituting v0 ( x, t ) 
w0 ( x, t )  q1 ( x)  q2 ( x)t into equation (3.7.b) one can get:
 2 v1 ( x, t )
 q ( x)  tq 2( x)  g ( x, t )
t 2 1
By integrating twice for both sides of the above differential equation with respect to t and by
v1 ( x, t )
using the initial conditions v1 (x,0)  0 and  0 one can obtain:
t t 0
t2 t3 t s
v1 ( x, t )  q1( x)  q2 ( x)    g ( x,  )dds.
2 6 00
By substituting v1 into equation (3.7.c) and by solving the resulting second order linear partial
v2 ( x, t )
differential equation together with the initial conditions v2 ( x,0)  0 and  0 one can
t t 0
get v 2 ( x, t ). In a similar manner one can get vi ( x, t ), i  3,4,.... By substituting vi ( x, t ), i  0,1,...
into equation (3.6) one can get the approximated solution w of the nonlocal problem given by
equations (3.3). Therefore from equation (3.2):

u ( x, t )  w( x, t )  z ( x, t )   vi ( x, t )  z ( x, t ), (x, t)  
i 0
which is the solution of the original nonlocal problem given by equations (3.1).
To illustrate this method consider the following example
- 52 -
Example (3.1):
Consider the homogeneous wave equation:
 2 u ( x, t )  2 u ( x, t )
  0, 0  x  , 0  t 1 (3.8.a)
t 2 x 2
u ( x , 0 )  cos( x ), 0 x  (3.8.b)
u ( x, t )
  cos( x), 0 x  (3.8.c)
t t 0
u ( x, t )
 0, 0  t 1 (3.8.d)
x x0

 u ( x, t )dx  0, 0  t  1 (3.8.d)
0
It is easy to check that the compatibility conditions are satisfied for this nonlocal problem. We
use the homotopy perturbation method to solve this example. To do this, let
u( x, t )
v0 ( x, t )  u0 ( x, t )  u( x,0)  t  cos(x)  t cos(x). From equation (3.7.b) and by
t t 0
using the initial conditions:
v1 ( x, t )
 v1 ( x,0)  0 one can have:
t t 0
1 2 1
v1 ( x, t )  t cos(x)  t 3 cos(x).
2! 3!
- 53 -
Hence
 1 1 
v0 ( x, t )  v1 ( x, t )  1  t  t 2  t 3  cos( x).
 2 6 
From equation (3.7.c) one can get:
t 
 2 v1 ( x, s )
v 2 ( x, t )    dsd
00  x 2
1 1
 t 4 cos( x )  t 5 cos( x ).
4! 5!
and this implies:
2
 1 1 1 1 
u ( x, t )   vi ( x, t )  1  t  t 2  t 3  t 4  t 5  cos( x).
i 0  2 3! 4! 5! 
and by continuing in this manner one can have:

u ( x, t )   vi ( x, t )  e t cos( x).
i 0
which is the exact solution of the nonlocal problem given by equations (3.8).
Example (3.2):
Consider the one-dimensional non-homogeneous wave equation:
 2 u ( x, t )  2 u ( x, t ) 
   x sin( t )  4e  2 x , 0  x  , 0  t  1 (3.9.a)
t 2
x 2
2

u ( x ,0 )  e  2 x , 0x (3.9.b)
2
- 54 -
u ( x, t ) 
 x, 0x (3.9.c)
t t 0 2
u ( x, t )
 sin(t )  2, 0  t 1 (3.9.d)
x x 0

2 1 1 1
 u ( x, t )dx  8 
2
sin(t )  e   , 0  t 1 (3.9.d)
0 2 2
It is easy to check that the compatibility conditions are satisfied for this nonlocal problem. We
use the homotopy perturbation method to solve this example. To do this, consider the
transformation given by equation (3.2). In this case:
1 1 1 1 
z ( x, t )  (sin(t )  2)( x   )  2(  2 sin(t )e  )e   1
4  8 2 
Therefore the nonlocal problem given by equations (3.9) is transformed to the one-dimensional
non-homogeneous wave equation
 2 w( x, t )  2 w( x, t ) 
  4e  2 x , 0 x , 0  t  1,
t 2
x 2
2
w( x,0)  e 2 x  2 x 

2

1

1  e ,

0x

2
,
w( x, t ) 
 0, 0x ,
t t 0 2
- 55 -
w( x, t )
 0, 0  t 1
x x 0

2
 w( x, t )dx  0, 0  t  1.
0
To solve this nonlocal problem by using the homotopy perturbation method, let
v0 ( x, t )  w0 ( x, t )  q1 ( x )  q 2 ( x )t

1
2

2e 2 x  4x   2  2e   2 
t2 t3 t s
 
v1 ( x, t )  q1 ( x)  q2 ( x)    g ( x,  )dds
2 6 00
0
Thus
vi ( x, t )  0, i  2,3,....
Therefore
w( x, t )  w0 ( x, t )

1
2

2e 2 x  4x   2  2e   2 . 
which is the exact solution of the above nonlocal problem.
Hence
u ( x, t )  w0 ( x, t )  z ( x, t )
 e -2x  x sin( x ).
which is the exact solution of the original nonlocal problem.

- 56 -
3.2 Solutions of the Hyperbolic Integro-Differential Equations with Non-Homogeneous
Neumann and Nonlocal Conditions:
Consider the hyperbolic integro-differential equation:
 2 u ( x, t )  2 u ( x, t ) t
  cu ( x, t )   k (t , s)u ( x, s)ds  f ( x, t ), (x, t)   (3.10.a)
t 2 x 2 0
u ( x,0)  r1 ( x ), 0x (3.10.b)
u ( x , t )
 r2 ( x ), 0 x (3.10.c)
t t  0
u ( x, t )
  (t ), 0t T (3.10.d)
x
 u ( x, t )dx   (t ), 0tT (3.10.e)

0
Where c is a known constant, f is a known function of x and t,   ( x, t ) 0  x  ,0  t  T ,
r1 , r2 , and  are given functions that must satisfy the previous compatibility conditions.
To solve this nonlocal problem by the homotopy perturbation method, we first transform
this nonlocal problem into another nonlocal problem, but with homogeneous Neumann condition
and homogeneous nonlocal condition. To do this we use the transformation given by equation
(3.2). Therefore the nonlocal problem given by equations (3.10) is transformed to the hyperbolic
integro-differential equation:
- 57 -
 w( x, t )  w( x, t )
2 2 t
  cw( x, t )   k (t , s ) w( x, s )ds  g ( x, t ), (x, t)   (3.11.a)
t 2 x 2 0
w( x,0)  q1 ( x ), 0  x  , (3.11.b)
w ( x , t )
 q 2 ( x ), 0  x  , (3.11.c)
t t  0
w( x, t )
 0, t  0 (3.11.d)
x x 0
 w( x, t )dx  0, t0 (3.11.e)

0
where
 2 z ( x, t ) t
g ( x, t )  f ( x, t ) 
t 2
 cz ( x , t )   k (t , s ) z ( x , s )ds ,
0
q1 ( x)  r1 ( x)  z ( x,0)
and
z ( x, t )
q2 ( x)  r2 ( x)  .
t t 0
To solve this nonlocal problem by the homotopy perturbation method, we construct a homotopy
v :   [0,1] 
 which satisfies:
- 58 -
 v( x, t , p )  w0 ( x, t )
2 2
 w0 ( x, t )
2
H (v, p )    p 
t 2 t 2 t 2
  2 v ( x, t , p ) t

p 
x
 cv ( x , t , p )   k (t , s)v( x, s, p )ds  g ( x, t )  0
 
2
0
(3.12)
where p  [0,1] ,  represents the set of all real numbers and w 0 is the initial approximation to
the solution of equation (3.11.a) which satisfies the initial condition, the Neumann condition and
the nonlocal condition given by equations (3.11.b)-(3.11.d).
 2 v( x, t ,0)  2 w0 ( x, t )
H ( v ,0 )  
t 2 t 2
 2 v( x, t ,1)  2 v( x, t ,1) t
H (v,1) 
t 2

x 2
 cv ( x , t ,1)   k (t , s)v( x, s,1)ds  g ( x, t )  0
0
Next, we assume that the solution of equation (3.11) can be expressed as in equation (3.5).
Therefore the approximated solution of the nonlocal problem given by equations (3.11) is given
by equation (3.6).
can get:
 2 vi ( x, t )  2 w0 ( x, t )

 2 w0 ( x, t )
H (v, p)   p i
 p 
i 0 t 2 t 2 t 2
   2 vi ( x, t )  t 

p   p i  c  p i
v ( x, t )   k (t , s )  p i vi ( x, s)ds  g ( x, t )  0
 i 0 x 2
i 0 0 i 0 
 2 v 0 ( x , t )  2 w0 ( x , t )
p0 :  0 (3.13.a)
t 2 t 2
- 59 -
 v1 ( x, t )
2
 w0 ( x, t )
2
 v 0 ( x, t )
2 t
1
p :    cv0 ( x, t )   k (t , s )v0 ( x, s )  g ( x, t )  0 (3.13.b)
t 2 t 2 x 2 0
 2 v j ( x, t )  2 v j 1 ( x, t ) t
j
p :   cv j 1 ( x, t )   k (t , s )v j 1 ( x, s )ds  0, j  2,3,... (3.13.c)
t 2 x 2 0
Similar to the previous, we take v0 ( x, t )  w0 ( x, t )  q1 ( x )  q2 ( x )t. In this case equation (3.13.a)
is automatically satisfied. By substituting it into equation (3.13.b) one can get:
 2 v1 ( x, t )
 q1( x)  q2 ( x)t  cq1 ( x)  q2 ( x)t  
t 2
t
 k (t , s)q1 ( x)  q2 ( x)sds  g ( x, t )
0
By integrating twice for both sides of the above differential equation with respect to t and by
v1 ( x, t )
using the initial conditions v1 (x,0)  0 and  0 one can obtain:
t t 0
t2 t3 1
v1 ( x, t )  q1( x)  q2 ( x)  cq1 ( x) t 2 
2 6 2
3 t2  t s
cq2 ( x)    k ( , s )q1 ( x)  q2 ( x) s dsddt 2    g ( x,  )dds
t
6 00 00
In a similar manner one can get vi (x, t), i  2,3,.... By substituting vi (x, t), i  0,1,... into
equation (3.6) one can get the approximated solution w of the nonlocal problem given by
equations (3.11). Therefore from equation (3.2):

u ( x, t )  w( x, t )  z ( x, t )   vi ( x, t )  z ( x, t ), (x, t)  
i 0
- 60 -
To illustrate this method consider the following example.
Example (3.3):
Consider the hyperbolic integro-differential equation:
u 2 ( x, t )  2u ( x, t ) t
1 34 1 33
  3u ( x , t )   (t  s 2
)u ( x , s )ds  6 xt  3tx 3
 xt  xt ,
t 2 x 2 0 4 2
0  x  1, 0  t  1.
(3.14.a)
u ( x,0)  0, 0  x  1, (3.14.b)
u ( x, t )
 x3 , 0  x  1, (3.14.c)
t t 0
u ( x, t )
 0, 0  t 1 (3.14.d)
x x 0
and the non-homogenous nonlocal condition:
1
1
 u ( x, t )dx  4 t , 0  t 1 (3.14.e)
0
We use the homotopy perturbation method to solve this example. To do this, we transform this
nonlocal problem into one but with homogeneous nonlocal condition. To do this, consider the
transformation given by equation (3.2). In this case:
1
z ( x, t )  t
4
- 61 -
and the nonlocal problem given by equations (3.14) consisting of the hyperbolic integro-
differential equation
 2 w( x, t )  2 w( x, t ) 1
t 2

x 2
 3 w( x , t )   (t  s 2 ) w( x, s )ds  6 xt  3tx 3 
0
1 3 4 1 3 3 3 1 1
x t  x t  t  t4  t3, 0  x  1, 0  t  1,
4 2 4 16 8
w( x,0)  0 0  x  1,
w( x, t ) 1
 x3  , 0  x  1,
t t 0 4
w( x, t )
 0, 0  t 1
x x0
 w( x, t )dx  0, 0  t  1.
0
To solve this nonlocal problem by using the ho motopy perturbation method, let
1
v0 ( x, t )  w0 ( x, t )  ( x 3  )t
4
 2 v1 ( x, t )
0
t 2
therefore
vi ( x, t )  0, i  1,2,...
- 62 -
and this implies that
1
w( x, t )  w0 ( x, t )  v0 ( x, t )  ( x 3  )t.
4
which is the exact solution of the above nonlocal problem.
Hence
u ( x, t )  w( x, t )  z ( x, t )
 tx 3 .
which is the exact solution of the original nonlocal problem.
- 63 -
Chapter Four
Solution of Some Real Life
Applications Via the Homotopy
Perturbation Method
Chapter Four Solution of Some Real Life Applications Via the Homotopy Perturbation Method
Introduction:
In this chapter we use the homotopy perturbation method for solving some real life
applications, namely advection-diffusion problems, gas dynamics equation and the
ground-water level equation.
This chapter consists of four sections:
In section one and two, we solved advection-diffusion problems and gas dynamics
via the homotopy perturbation method.
In section three, we present the homotopy perturbation method for solving the ground-
water level problem.
4.1 Advection-Diffusion Problems:
Problems involving diffusion-advection equations arise in many domains of science.
There are several methods for solving these quations, like the differential transform
method, [34]. In this section, we use the homotopy perturbation method to solve the
advection-diffusion problem that consists of the advection-diffusion equation:
u ( x, t )  2 u ( x, t ) u ( x, t )
   s ( x), (x, t)   (4.1.a)
t x 2
x
together with initial condition:
u ( x,0)  r ( x), 0  x  , (4.1.b)
and the boundary conditions:
u (0, t )  f (t ), t  0, (4.1.c)
u (, t )  g (t ), t  0, (4.1.d)
- 64 -
where the first two terms on the right hand side represent different physical processes:
 2 u ( x, t ) u ( x, t )
 corresponds to normal diffusion while  describes advection which is
x 2 x
why the equation is also known as the advection-diffusion equation. Further u is the
variable of interest (species concentration for mass transfer). And ,  are non-negative
real numbers where  is the diffusivity for species or heat transfer and  is the velocity,
u is a known function of x only and   {(x, t) 0  x  , t>0}, r, f and g are given
functions that must satisfy the compatibility conditions:
r(0)  f (0)
and
r()  g(0).
To solve this problem by the homotopy perturbation method, we first transform this
problem into another problem, but with homogeneous boundary conditions. To do this we
use the transformation:
w(x, t)  u(x, t)  z(x, t), (x,t)   (4.2)
where
1
z(x, t)  f (t)  f (t)  g(t)  x

Then
u(x, t) w(x, t) z(x, t)

 
t t t
and
 2 u(x, t)  2 w(x, t)
 .
x 2 x 2
- 65 -
Therefore the problem given by equations (4.1) is transformed to the partial differential
equation:
w(x, t)  2 w(x, t) w(x, t)

    g(x, t), (x,t)  (4.3.a)
t x 2 x
w(x,0)  q(x), 0  x  , (4.3.b)
and the homogeneous boundary conditions:
w(0, t)  0, t  0 (4.3.c)
w(, t)  0, t  0 (4.3.d)
z(x, t) z(x, t)
where g(x, t)  s(x)    and q(x)  r(x)  z(x,0).
x t
To solve this problem by the homotopy perturbation method, we rewrite equation
(4.3.a) as
A(w)  g(x, t)  0
w 2w w
where A(w)   2  . Then the operator A can be divided into two parts L
t x x
and N such that equation (4.3.a) becomes:
L(w)  N(w)  g(x, t)  0
 2 
where L  and N   2   .
t x x
According to [13], we can construct a homotopy v :   [0,1] 

 which satisfies
v(x, t,p) w 0 (x, t) w (x, t)

H(v,p)   p 0 
t t t
  2 v(x, t,p) v(x, t,p) 
p      g(x, t) 0
 x 2
x 
- 66 -
(4.4)
where p  [0,1] is an embedding parameter and w o is the initial approximation to the
solution of equation (4.3.a) which satisfies the initial condition and the boundary
conditions given by equations (4.3.b)-(4.3.d).
Next, we assume that the solution of equation (4.4) can be expressed as

v(x, t,p)   pi vi (x, t) (4.5)
i 0
Therefore the approximated solution of the problem given by equations (4.3) can be
obtained as follows:

w(x, t)  lim v(x, t,p)   vi (x, t) (4.6)
p1
i 0
By substituting the approximated solution given by equation (4.5) into equation
(4.4) one can get:


vi (x, t) w 0 (x, t) w (x, t)
H(v,p)   pi  p 0 
i 0 t t x
 
 2 vi (x, t) 
i v i (x, t) 
p    pi    p  g(x, t) 0
 i 0  x 2
i 0  x 
v0 (x, t) w 0 (x, t)

p0 :  0 (4.7.a)
t t
v1 (x, t) w 0 (x, t)  2 v 0 (x, t) v 0 (x, t)

1
p :      g(x, t)  0 (4.7.b)
t t x 2 x
and in general
v j (x, t)  2 v j1 (x, t) v j (x, t)

j
p :    0, j=2,3,... (4.7.c)
t x 2
x
- 67 -
For simplicity, we take v0 (x, t)  w 0 (x, t) . In this case equation (4.7.a) is automatically
satisfied. Let w 0 (x, t)  q(x) then
w 0 (x,0)  q(x), 0  x  , w o (0, t)  q(0)  r(0)  z(0,0)  r(0)  f (0)  0, t  0
and
w o (, t)  q()  r()  z(,0)  g(0)  g(0)  0, t  0
Therefore w o satisfies the initial and the boundary conditions given by equations (4.3.b)-
(4.3.d). Therefore by substituting t=0 in equation (4,6) one can have:

w(x,0)   vi (x,0)
i 0
But w 0 (x,0)  q(x) and w(x,0)  q(x) , hence vi (x,0)  0, i=1,2,.... By substituting
v 0 (x,0)  w 0 (x, t)  q(x) into equation (4.7.b) one can get:
v1 (x, t)
 q(x)   q(x)  g(x, t)
t
By integrating both sides of the above differential equation and by using the initial
condition v1 (x,0)  0 one can obtain:
t
v1 (x, t)   q(x)  q(x)  t   g(x, )d
0
By substituting v1 into equation (4.7.c) and by solving the resulting first order linear
partial differential equation together with the initial condition v 2 (x,0)  0 one can get
v 2 (x, t) . In a similar manner one can get vi (x, t), i  3,4,.... By substituting
vi (x, t), i  0, 1,... into equation (4.6) one can get the approximated solution of the
problem given by equations (4.3). Therefore from equation (4.2):
- 68 -

u(x, t)  w(x, t)  z(x, t)   v i (x, t) z(x, t), (x,t)  
i 1
which is the solution of the advection-diffusion problem given by equations (4.1).
4.2 Gas Dynamics problem:
Consider the nonlinear non-homogeneous gas dynamic equation:
u ( x, t ) u ( x, t )
 u ( x, t )  u ( x, t )1  u ( x, t )  f ( x, t ), (x, t)   (4.8.a)
t x
u ( x,0)  r ( x), 0  x  1 (4.8.b)
where f is a known function of x and t   {( x, t ) | 0  x  1, t  0}
In [22], they use the homotopy perturbation method for solving the homogeneous
gas dynamic equation in case r ( x)  e  x , 0  x  1. Here we use the same method to solve
the non-homogeneous gas dynamic equation for any choice of the initial condition. To do
this, we construct a homotopy v :   [0,1] 

v( x, t , p ) u 0 ( x, t ) u ( x, t )
H (v, p )   p 0 
t t t
 v( x, t , p ) 
p   v ( x, t , p )  v( x, t , p )[1  v( x, t , p )]  0 (4.9)
 x 
where p  [0,1] and u o is the initial approximation to the solution of equation (4.8.a)
which satisfies the initial condition given by equation (4.8.b).
v( x, t ,0) u 0 ( x, t )
H ( x ,0 )   0
t t
- 69 -
v( x, t ,1) v( x, t ,1)
H (v,1)   v( x, t ,1)  v( x, t ,1)[1  v( x, t ,1)]  0.
t t
Next, we assume that the solution of equation (4.9) can be expressed as in
equation (3.3). By substituting this approximated solution into equation (4.9) one can get:
 vi ( x, t ) u 0 ( x, t ) u ( x, t )
H (v, p )   p i  p 0 
i 0 t t t
   v ( x, t )  i  

p   p i vi ( x, t ) p i i   p vi ( x, t )1   p i vi ( x, t )   0`
 i 0 i 0 x i 0  i 0 
v0 ( x, t ) u 0 ( x, t )
p0 :  0 (4.10.a)
t t
v1 ( x, t ) u 0 ( x, t ) v ( x, t )
p1 :   v0 ( x, t ) 0  v0 ( x, t )1  v0 ( x, t )  0 (4.10.b)
t t x
v2 ( x, t ) v ( x, t ) v ( x, t )
p2 :  v0 ( x, t ) 1  v1 ( x, t ) 0  2v0 ( x, t )v1 ( x, t )  v1 ( x, t )  0 (4.10.c)
t t x
For simplicity, we take v0 ( x, t )  u 0 ( x, t ). In this case equation (4.10.a) is automatically
satisfied. Let u 0 ( x, t )  r ( x) then u 0 ( x,0)  r ( x), 0  x  1. Therefore u 0 satisfies the the
initial condition given by equation (4.8.a). Thus:

u ( x,0)   vi ( x,0)
i 0
But v0 ( x,0)  u ( x,0)  r ( x) hence v i ( x ,0)  0, i  1,2,.... By substituting v0 ( x,0)  u 0 ( x, t )
 r (x) into equation (4.10.b) one can get:
v1 ( x , t )   r ( x ) r ( x )  r ( x )  r ( x ) 
2
t
- 70 -
By substituting v1 into equation (4.10.c) and by solving the resulting first order linear
partial differential equation together with the initial condition v2 ( x, t )  0 one can get

v2 ( x, t ). In a similar manner one can get vi ( x, t ), i  3,4,.... Thus u ( x, t )   vi ( x, t ) is the
i 0
approximated solution of the initial value problem given by equations (4.8).
4.3 The Ground Water Level problem:
Consider the linear partial differential equation:
 2 ( x, t ) 1    
  x   h( x, t ), (x, t)  (0, b)  (0, T] (4.11.a)
t x x  x 
 ( x,0)  r ( x), 0  x  b (4.11.b)
 ( x, t )
 p ( x), 0  x  b (4.11.c)
t t 0
the non-homogeneous Dirichlet condition:
 (b, t )   (t ), 0  t  T (4.11.d)
1b
  ( x, t )dx   (t ), 0  t  T
b0
(4.11.e)
where  is the ground water level,  (t ) is the mean value of  at time t and h is a known
function of x and t and r , p, and  are given functions that must satisfy the
compatibility conditions:
r (b )   (0),
- 71 -
p (b )   (0),
1b
 r ( x)dx   (0),
b0
and
1b
 p( x)dx   (0).
b0
To solve this nonlocal problem by the homotopy perturbation method, we first transform
this nonlocal problem into another nonlocal problem, but with homogeneous Dirichlet and
nonlocal conditions. To do this we use the transformation that appeared in [5]:
w( x, t )   ( x, t )  z ( x, t ), (x, t)  (0, b)  (0, T ] (4.12)
where z ( x, t )  2  (t )   (t )   (t )   (t ). Then the nonlocal problem given by

2x
b
equations (4.11) is transformed to the one-dimensional non-homogeneous linear partial
differential equation:
 2 w( x, t ) 1   w( x, t ) 
 x   g ( x, t ), (x, t)  (0,1)  (0, T] (4.13.a)
t 2 x x  x 
w( x,0)  q1 ( x ), 0  x  1 (4.13.b)
w( x, t )
 q2 ( x), 0  x  1 (4.13.c)
t t 0
and the homogeneous Dirichlet conditions:
w(b, t )  0, 0  t  T (4.13.d)
and
- 72 -
b
1
b0
 w( x, t )dx  0, 0  t  T (4.13.e)
 2 z ( x, t ) 1   z ( x, t ) 
where g ( x, t )  h ( x, t )   x  and q1 ( x)  r ( x)  z ( x,0) and
t 2 x x  x 
z ( x, t )
q2 ( x)  p( x)  .
t t 0
To solve this nonlocal problem by the homotopy perturbation method, we can
construct a homotopy v : (0,1)  (0, T ]  [0,1] 

 2 v( x, t , p )  2 w0 ( x, t )  2 w0 ( x, t )  1   w( x, t )  
H (v, p )   p  p  x   g ( x, t )   0
t 2 t 2 t 2  x x  x  
(4.14)
where p  [0,1] and w 0 is the initial approximation to the solution of equation (4.13.a)
which satisfies the initial condition and the nonlocal conditions given by equations
(4.13.b)-(4.13.d).
 2 v( x, t ,0)  2 w0 ( x, t ,0)
H ( v ,0 )   0
t 2 t 2
 2 v( x, t ,1) 1   v( x, t ,1) 

H (v,1)   x   g ( x , t )  0.
t 2 x x  x 
Next, we assume that the solution of equation (4.14) can be expressed as:

v ( x , t )   p i vi ( x , t ) (4.15)
i 0
Therefore the approximated solution of the nonlocal problem given by equations (4.13) is
given by:
- 73 -

w( x, t )   vi ( x, t ) (4.16)
i 0
By substituting the approximated solution given by equation (4.15) into equation (4.14)
one can get:
  2 vi ( x, t )  2 w0 ( x, t )  2 w0 ( x, t )  1    i vi ( x, t )  
H (v, p)   p i   p  p    x  p   g ( x, t )   0.
i 0 t 2 t 2 t 2  x x  i  0 x  
Then by equating the terms with identical powers of p one can obtain:
 2 v0 ( x, t )  2 w0 ( x, t )
0
p :  0 (4.17.a)
t 2 t 2
 2 v1 ( x, t )  2 w0 ( x, t ) 1   v0 ( x, t ) 
1
p :   x   g ( x, t )  0 (4.17.b)
t 2 t 2 x x  x 
 2 v j ( x, t ) 1   v j 1 ( x, t ) 
j
p :  x   0, j  2,3,... (4.17.c)
t 2 x x  x 
For simplicity, we take v0 ( x, t )  w0 ( x, t ). In this case equation (4.17.a) is automatically
satisfied. Let w0 ( x, t )  q1 ( x)  q2 ( x)t then
w0 ( x,0)  q1 ( x), 0  x  b,
w0 ( x, t )
 q2 ( x), 0  x  b,
t t 0
w0 (b, t )  q1 (b)  q2 (b)t

z ( x, t )
 r (b)  z (b,0)  p (b)t  t
t tx0b
  (0)   (0)   (0)t   (0)t
 0, 0tT
and
- 74 -
b b
w0 ( x, t )dx   q1 ( x )  q 2 ( x )t dx
1 1
b0
 b0
1b 1b 1 b t b z ( x , t )
 
b0
r ( x ) dx  
b0
z ( x , 0 ) dx  t
b 0
 p ( x ) dx  
b 0 t t  0
dx
1b 1 b 2x
  (0)   2  (0)   (0)dx    (0)   (0)dx  t (0)
b0 b0 b
1 b 1 b 2x
 t  2  (0)   (0)dx  t   (0)   (0)dx
b 0 b 0 b
  (0)  2  (0)   (0)   (0)   (0)  t (0)  t 2  (0)   (0)  t  (0)   (0)
 0, 0  t  T.
Therefore w o satisfies the initial condition and the nonlocal conditions given by equations
(4.13.b)-(4.13.d). By substituting v0 ( x,0)  w0 ( x, t )  q ( x) into equation (4.13.b) and by
using the initial condition v1 ( x,0)  0 one can get:
t
v1 ( x, t )  q( x)t   g ( x, )d
0
From equation (4.13.c) and by using the initial condition vi (x,0)  0, i=2,3,... one can
have:
t  2 v j 1 ( x, )
v j ( x, t )   d , j  2,3,...
0 x 2
Therefore from equation (4.12):

u ( x, t )  w( x, t )  z ( x, t )   vi ( x, t )  z ( x, t ), (x, t)  (0,1)  (0, T ]
i 1
- 75 -
Conclusions and Recommendation
From the present study, we can conclude the following:
(1)The homotopy perturbation method can also be used to solve the linear Volttera
integral equation of the first kind by transferring it into an equivalent integral equation
of the second kind.
(2) The homotopy perturbation method for solving linear integral equations of the second
order is precisely the method of successive in case the initial approximation is
u 0 ( x)  f ( x) instead of u 0 ( x)  0.
(3) The homotopy perturbation method for solving any initial or boundary value problems
requires the initial approximation to the solution of these problems must satisfy the
initial or boundary conditions associated with these problems.
(4) The homotopy perturbation method can be also used to solve systems of differential,
integral and integro-differential equations.
Also, we recommend the following for future work:-
(1) Discuss the convergence of the homotopy perturbation method for the prescribed non-
local problems.
(2) Use the homotopy analysis method to solve the nonlocal problems.
(3) Solve the fuzzy integro-differential equations via the homotopy perturbation method.
- 76 -
References
[1] Allahviranloo T., Khezerloo M., Ghanbari M. and Khezerloo S., "The Homotopy
Perturbation Method for Solving Fuzzy Volttera Integral Equations", Int. J. of
Comp. Co., Vol.8, No.2, pp.31-37, 2010.
[2] Abbasbandy S., "Numerical Solutions of the Integral Equations", Appl. Math.,
Vol.173, pp.493-500, 2006.
[3] Abasbandy A. and Roohani H. G., "The He’s Variational Iteration Method for
Solving the Integro- Differential Parabolic Problem with Integral Conditions"
Appl. and Appl. Math. In., Special Issue, No.1, pp.12-23, 2010.
[4] Behrouz R., "Application of He’s Homotopy Perturbation Method and Variational
Iteration Method for Nonlinear Partial Integro-Differential Equations ", World
Appl.Sc. J., Vol.7, No.4, pp.399-404, 2009.
[5] Bouziani A., "On Initial Boundary Value Problem with Dirichlet Integral Conditions
for a Hyperbolic Equation with the Bessel Operator", J. Appl. Math., Vol.10,
pp.487-502, 2003.
[6] Beilin S., "Existence of Solutions for One-Dimensional Wave Equations with
Nonlocal Conditions", Elect. J. Diff. Eq., Vol.76, pp.1-8, 2002.
[7] Cannon J. and Lin Y., "A Galerkin Procedure for Diffusion Equations with Boundary
Integral Conditions", Inter. J. Engng. Sc., Vol.28, pp.579-587, 1990.
[8] Carlson D., "Linear Thermoelasticity", Encyclopedia of Physics, Springer, Berlin,
1972.
- 77 -
[9] Fatemeh Sh., Mehdi D., "Solution of Delay Differential Equations via a Homotopy
Perturbation Method ", J. Appl. Sc., Vol.8, No.7, pp.1256-1261, 2008.
[10]El Qarnia H., "Application of Homotopy Perturbation Method to Non-
Homogeneous Parabolic Partial and Nonlinear Differential Equations" Worl. J.
Model. Simul., Vol.5, No.3, pp.225-231, 2009.
[11] Ghorbali A. R., Rezania A., Ganji D.D.and Bararnia H., "Application on Homotopy
Perturbation and Variational Iteration Methods for Heat Equation", Aust. J. Bas.
Appl. Sc., Vol.3, No.3, pp.1863-1874, 2009.
[12] Hwang C., Lin S., Shen L., "Effects of Wall Conduction and Interface Thermal
Resistance on the Phase Change Problem", Int. J. Heat. Mass Trans., Vol.37, No.13,
pp.1844-1855, 1994.
[13] He J. H., "Homotopy Perturbation Technique", Comput. Math. Appl. Mech. Engrg.,
Vol.178, pp.257-262, 1999.
[14] He J. H., "Homotopy Perturbation Method for Solving Nonlinear Problem", Appl.
Math. Comput., Vol.135, pp.73-79, 2003.
[15] He J. H., "A coupling Method of A Homotopy Technique and A Perturbation
Technique for Non-Linear Problems", Int. Non-Linear Mech, Vol.35, pp.37-43,
2000.
[16] He J. H., "A New Non-linear Analytical Technique", Appl. Math. Comp., Vol.135,
pp.73-79, 2000.
[17] He J. H., "The Homotopy Perturbation Method for Solving Non-Linear Oscillators
with Discontinuities", Appl. Math. Comp., Vol.151, pp.287-292, 2004.
- 78 -
[18] He J. H.,"The Homotopy Perturbation Method for Solving Nonlinear Wave
Equation", Choas. Solit. and Fract., Vol.26, pp.695-700, 2005.
[19] He J. H. and Li-Naz., " Homotopy Perturbation Method for the Solution of The
Electrostatic Potential Differential Equation", Math. Probl. Engin., Vol.10, pp.1-6,
2006.
[20] Han M., Huang P. "Perturbation Solutions of Planar Diffusion-Controlled Moving-
Boundary Problems". Int, J. Heat. Mass Transf., Vol.18, pp.689-695, 1975.
[21] Jafari H., Alipour M. and Tajadodi H. , "Convergence of Homotopy Perturbation
Method for Solving Integral Equations", Thai J. Math., Vol.8, No.3, pp.511-520,
2010.
[22] Jafari H., Zabihi v and Saidy M., "Application of Homotopy Perturbation Method
For Solving Gas Dynamics Equation", Appl. Math. Sc., Vol.2, No.48, pp.2393-
2396, 2008.
[23] Jazbi B. and Moini. M. "Application of He’s homotopy Perturbation Method for
Solving Schrodinger Equation". Iranian J. Math. Sc. Inf., Vol.3, No.2, pp.13-19,
2008.
[24] Javidi, M. and Golbabai A., "A Numerical Solution for Solving System of
Fredholm Integral Equations by Using Homotopy Perturbation Method", Appl.
Math. Comput., Vol.189, pp.1921-1928, 2007.
[25] Karakostas G. L. and Tsamatos P. Ch. "Positive Solutions for a Nonlocal
Boundary-Value Problem with Increasing Response", Elect. J. Diff. Eq.,Vol.2000,
No.37, pp.1-8, 2000.
- 79 -
[26] Khaleel A., "Homotopy Perturbation Method for Solving Spaecial Type of
Nonlinear Fredholm Integor-Differential Equations", J. Al-Nahrain University,Vol.
13, No.2, pp.219-224, 2010.
[27] Khaleel A., "Homotopy Perturbation Method for Solving Some Types of Nonlocal
Problem With Some Real Life Applications" to be appear in J. AL-Nahrain
University
[28] Khamas A., "Some Approximation Methods for Solving Linear System of Integr-
Differential Equations", Msc. Thesis, College of Education, Ibn Al-Hatham,
University of Baghdad, 2009.
[29] Lin J., "Homotopy Perturbation Technique for Solving Partial Differential
Equations with Variable Coefficients", Int. J. Contemp. Math. Sc., Vol.3, No. 28,
pp.1395-1407, 2008
[30] Mohammad A., "A Class of Hyperbolic Equations with Nonlocal Conditions", Int.
J. of Math. Analy., Vol.2, No.10, pp.491-498, 2008.
[31] Max D. L., "Fundamental Concepts of Modern Mathematics", Addision-Wesley,
1970.
[32] Massey W. S., "Algebric Topology", Harcourt, Brace Worl. Inc., 1967.
[33] Mashallah M. and Mohammad S., "The Homotopy Perturbation Method for Solving
Fuzzy Integral Equations", The J. of Math. Comput. Sc.,Vol.1, No.4, pp.377-385,
2010.
[34] Patrıcio M. F. and P. M. Rosa, "The Differential Trnasform Method for Solving
Reaction-Diffusion Equations", Int. J. Comput. Math. Sc., Vol.1, No.4, pp.258-262,
2007.
- 80 -
[35] Roozi A., Alibeiki E., Hosseini S.S., Shaﬁof S.M. and Ebrahimi M." Homotopy
Perturbation Method for Special Nonlinear Partial Diﬀerential Equations", J. King
Saud University, Vol.23, pp.99-103, 2011.
[36] Ruyun M. "A Survey On Nonlocal Boundary Value Problems". Appl. Math., Vol.7,
pp.257-279, 2007.
[37] Saeed R. K., "Homotopy Perturbation Method for Solving System of Non-Linear
Fredholm Integral Equations of The Second Kind", J. Appl. Scien. Res., Vol.4,
No.10, pp.1166-1173, 2008.
[38] Yu-Xi W., Hua-You S., and Lu-Feng M., "Homotopy Perturbation Method for
Solving Reaction-Diffusion Equations", Math. Probl. Eng., Vol.10, pp.1-5, 2008.
- 81 -
‫اﻟﻤﺴﺘﺨﻠﺺ‬
‫ﻤﺤﻠﻴـﺔ‬ ‫ﻟﻤﺘﺨﻠﺨﻞ‬
‫ﻟﻼﻣﺤﻠﻴﺔ ‪.‬‬
‫‪:‬‬
‫)‪(1‬‬
‫‪.‬‬
‫ﻤﺘﺨﻠﺨـﻞ‬ ‫)‪(2‬‬
‫‪.‬‬ ‫‪-‬‬
‫)‪(3‬‬
‫ﻟﺤﻠﻬﺎ‪.‬‬
‫)‪(4‬‬
‫ﻣﺴـﺄﻟﺔ ﻣ‬ ‫‪،‬‬
‫‪.‬‬
‫ﻗﺴﻢ‬
‫ﺣﻮل ﻃﺮﯾﻘﺔ اﻟﮭﻮﻣﻮﺗﻮﺑﻲ اﻟﻤﺘﺨﻠﺨﻞ‬

‫وﺗﻄﺒﯿﻘﺎﺗﮭﺎ‬
‫رﺳﺎﻟﺔ‬
‫‪-‬‬ ‫ﻣﻘﺪﻣﺔ‬
‫ﻓﻲ‬
‫ﻣﻦ ﻗﺒﻞ‬
‫‪(2008 ،‬‬ ‫‪،‬‬ ‫)‬
‫‪. .‬‬ ‫‪. . .‬‬
‫‪1432‬‬ ‫رﺑﯿﻊ اﻟﺜﺎﻧﻲ‬

‫‪2011‬‬ ‫آذار‬

On The Homotopy Perturbation Method and Its Applications: Farah Lateef Joey

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

On The Homotopy Perturbation Method and Its Applications: Farah Lateef Joey

Uploaded by

Copyright:

Available Formats

Republic of Iraq

Ministry of Higher Education

On the Homotopy Perturbation

Rabee Al-Thanee 1432

‫اﻟﻰ ﻣﺜﻠﻲ اﻻﻋﻠﻰ اﻟﺬي ﻛﺎن ﯾﺸﺠﻌﻨﻲ دوﻣﺎ‬

‫اﻟﻰ ﺷﻤﺲ اﺷﺮﻗﺖ ﻓﻲ وﺟﮭﻲ وﺗﺪﻓﻌﻨﻲ‬

‫واﻟﻰ ﻛﻞ ﻣﻦ ﯾﻌﺘﺒﺮ اﻟﺮﯾﺎﺿﯿﺎت ﻃﺮﯾﻘﺎً ﻓﻲ اﻟﺤﯿــــﺎة‬

Thanks to Allah for helping me to complete my thesis

(1) Give some basic concepts of the homotopy perturbation method.

and integro-differential equations.

-water level problem.

Chapter Two: The Homotopy Perturbation Method for

Chapter Four: Solutions of Some Real Life Applications Via

Conclusions and Recommendations………………………….76

converge to the accurate solution, [13].

The homotopy perturbation method, presents some advantages: obtaining exact

mechanics and heat transfer, [10].

Schrodinger equation, [23].

Mashallah M. and Mohammad S. in 2010, studied the homotopy perturbation method

hyperbolic partial differential equations of one and two dimensions, [35].

various field of mathematical physics and in other fields, [3].

One-Dimensional Wave equations with Nonlocal conditions, [6], Ruyun M. in 2007,

This thesis consists of four chapters:

obtained by means of the homotopy perturbation method.

dimensional wave and hyperbolic integro-differential equations with non-homogeneous

Neumann and nonlocal conditions, respectively.

approximated solution to a wide range of nonlinear problems in applied sciences. In this

converges rapidly to the solutions, [13].

This chapter consists of two sections:

1.1 Some Basic Concepts of the Homotopy Perturbation Method:

do this, we recall the following definition:

Definition (1.1.1), [32]:

Let X and Y be two topological spaces. Two continuous functions f : X 

In this case, H is said to be a homotopy.

Now, to illustrate this definition, consider the following examples:

function, then define H : X  [0, 1] 

Then H is a continuous function and

On the continuous functions  is an equivalence relation.

and this implies that f  f . Therefore  is a reflexive relation.

To prove  is a symmetric relation, let f  g , then there exists a continuous function

K ( x,0)  H ( x,1)  g ( x), x  X

K ( x,1)  H ( x,0)  f ( x), x  X

To prove  is a transitive relation, let f  g and g  w, then there exist continuous

lim L( x, p )  lim H ( x, p )  K ( x, p )  g ( x)

lim L( x, p )  lim H ( x, p )  K ( x, p )  g ( x)

Hence L is a continuous function. Therefore f  w. Hence  is an equivalence relation on

the set of all continuous functions.

Let X and Y be two topological spaces, let f :  

continuous functions. Define H :   [0, 1] 

Definition (1.1.5), [32]:

Let X and Y be two topological spaces and f : X 

equivalence class of f , denoted by [ f ] is defined by:

 Y be a continuous function and f  g 

and it is said to be a homotopy class of functions of f .

X into a topological space Y. Therefore [ f ]   .

Definition (1.1.7), [32]:

homotopic relative to A  X if there exists a continuous function H : X  [0,1] 

following non-linear equation:

A(u )  f ( x), x   (1.1)

operator. Therefore equation (1.1) can be rewritten as follows:

According to [13], we can construct a homotopy v :   [0,1] 

H (v, p)  (1  p)[ L(v)  L(u 0 )]  p[ A(v)  f ( x)]  0